Hennepin County Veterans Court Evaluation Prepared By€¦ · Matthew Johnson would personally like to thank the Hennepin CountyDistrict Court Research Division—Marcy Podkopacz,

Hennepin County Veterans Court Evaluation

Prepared By:

Matthew Johnson, Ph.D.

Research Analyst II

December 9, 2016

Fourth Judicial District of Minnesota ● Hennepin County, Research Division:

Marcy R. Podkopacz, Ph.D., Director

Jackie Braun-Lewis, M.P.P., Research Analyst I

Matthew A. Johnson, Ph.D., Research Analyst II

Dana Hurley Swayze, M.S.W., Research Analyst II

www.mncourts.gov/Find-Courts/Fourth-Judicial-District/Publications-and-Reports-Hennepin.aspx

i

Fourth Judicial District of Minnesota ● Hennepin County

Acknowledgements

Several individuals and organizations were indispensable for this evaluation. The Hennepin

County District Court team wishes to thank the St. Paul Veterans Benefit Administration, and

especially Pam Werdal and Nick Andro for helping to identify the comparison group of justice-

involved veterans who did not participate in the Veterans Court Program. David Holewinski, the

Veteran Justice Outreach Coordinator, was also vital in providing important information for this

evaluation. Lori Swenson and the Veterans Court probation officers from the Hennepin County

Department of Community Corrections and Rehabilitation provided extremely important

information for the Veterans Court participants evaluated in this study. Allison Holbrook provided

very helpful and timely information about the workings of the program and detailed information

on many participants. Milissa Salmonson’s consistent dedication to providing detail-oriented data

for all treatment court participants is invaluable. Veterans Court Presiding Judge Kerry Meyer was

also instrumental to this evaluation (in addition to the all of the previous Veterans Court Presiding

Judges). In addition, we would like to thank the remainder of the Veterans Court steering

committee and court team.

Matthew Johnson would personally like to thank the Hennepin County District Court Research

Division—Marcy Podkopacz, Jackie Braun-Lewis, and Dana Hurley Swayze—for helpful

comments and ideas of previous versions of this evaluation, as well as Tracy Loynachan, Kurt

Weyland, and Raul Madrid.

ii


Table of Contents

Acknowledgements ........................................................................................................................................... i

List of Tables and Figures ............................................................................................................................. iii

Executive Summary ........................................................................................................................................ vi

Introduction ...................................................................................................................................................... 1

Overview of the Hennepin County District Court Veterans Court Program ........................................ 2

The Veterans Court Program in Action: Review Hearings and the Veterans Court Teams .............. 10

Research Design ............................................................................................................................................. 12

Demographic Profile of Veterans Court Cohort ...................................................................................... 16

Evaluating Outcomes of Veterans Court Participants ............................................................................. 32

Recommendations ......................................................................................................................................... 64

Conclusion ...................................................................................................................................................... 67

References ....................................................................................................................................................... 71

iii


List of Tables and Figures

Table 1: Gender Breakdown of Veterans Court Participants .................................................................. 16

Table 2: Race/Ethnicity Breakdown of Veterans Court Participants ..................................................... 17

Figure 1: Veterans Court Success Rates, by Racial/Ethnic Group ......................................................... 18

Table 3: Age of Veterans Court Participants, Descriptive Statistics ....................................................... 18

Table 4: Employment Status at Veterans Court Entry ............................................................................. 19

Table 5: Education Status at Veterans Court Entry .................................................................................. 20

Table 6: Housing Status at Veterans Court Entry ..................................................................................... 20

Table 7: Charge Level at Disposition for Veterans Court Instant Offense ........................................... 21

Figure 2: Veterans Court Instant Offense by Type ................................................................................... 21

Table 8: Convictions before Entering Veterans Court ............................................................................. 23

Table 9: Months Spent in Veterans Court Program .................................................................................. 24

Table 10: Number of Veterans Court Review Hearings ........................................................................... 24

Figure 3: Branches of Military Service of Veterans Court Participants .................................................. 25

Table 11: Discharge Type of Veterans Court Participants ....................................................................... 26

Table 12: Rank at Discharge from Armed Forces* ................................................................................... 26

Table 13: Deployment History of All Veterans Court Participants ........................................................ 27

Table 14: Combat Deployment History of Veterans Court Participants (of the 86 Participants

Deployed) .................................................................................................................................. 27

Table 15: Instant Offense Charge Type of Comparison Group and Veterans Court Cohort ............ 29

Table 16: Racial Composition of the Comparison Group and Veterans Court Cohort ...................... 30

Table 17: Gender Breakdown of the Comparison Group and Veterans Court Cohort ...................... 30

Table 18: Prior Convictions of Comparison Group and Veterans Court .............................................. 31

Table 19: Average Age of Comparison Group and Veterans Court Participants ................................. 32

iv


Table 20: Percentage of Veterans Court Participants Subjected to In-Program Drug Testing by

DOCCR ..................................................................................................................................... 33

Table 21: In-Program Drug Test Information for Veterans Court Participants ................................... 34

Figure 4: Proportion of Veterans Court Participants Active in 4th Quarter 2012 Receiving

Drug/Alcohol Test, by Testing Site (N=61) ....................................................................... 36

Table 22: Mental Health Diagnosis of Veterans Court Participants at Program Entry and Exit ....... 38

Table 23: Treatment Outcomes for Veterans Court Participants (of the 106 Participants who

Underwent Treatment) ............................................................................................................ 39

Table 24: New Criminal Charges during Veterans Court ......................................................................... 39

Table 25: Number of Warrants Issued during Veterans Court ............................................................... 40

Table 26: Change in Veterans Administration Benefit Connectedness during Veterans Court ......... 41

Table 27: Prosocial Activities at Veterans Court Entry and Exit ............................................................ 42

Table 28: Prosocial Relationships at Veterans Court Entry and Exit ..................................................... 43

Table 29: Change in Number of Prosocial Activities from Veterans Court Entry to Exit* ............... 43

Table 30: Change in Number of Prosocial Relationships from Veterans Court Entry to Exit* ........ 44

Table 31: Mentor Program Participation during Veterans Court ............................................................ 44

Table 32: Housing Status at Veterans Court Entry and Exit ................................................................... 46

Figure 5: Housing Status Change from Veterans Court Start to Finish ................................................. 47

Table 33: Employment Status at Veterans Court Entry and Exit ........................................................... 48

Figure 6: Employment Status Change from Veterans Court Start to Finish......................................... 49

Table 34: Education Status at Veterans Court Entry and Exit ................................................................ 50

Figure 7: Education Status Change from Veterans Court Start to Finish ............................................. 51

Figure 8: Percent of Veterans Court and Comparison Group who Recidivated .................................. 54

Table 35: Recidivism Analysis between Veterans Court Cohort and Comparison Group ................. 55

Figure 9: Time to Recidivate for Veterans Court Participants and the Comparison Group .............. 56

v


Figure 10: Total Number of New Convictions, and New Convictions by Type .................................. 57

Table 36: Recidivism Analysis between Veterans Court Graduates and Comparison Group ............ 58

Figure 11: Recidivism among Veterans Court Graduates and Non-Completers .................................. 59

Table 37: Determinants of Veterans Court Success .................................................................................. 62

Table 38: Goal Scorecard for Hennepin County Veterans Court Evaluation ....................................... 68

vi


Executive Summary and Recommendations

This evaluation explores the outcomes of the Hennepin County District Court Veterans Court

program. The following are key findings and recommendations from the evaluation, which focuses

on 155 Veterans Court participants active in the program between 2010 and 2013 and completed the

program two years before June 2016.

Hennepin County District Court initiated its Veterans Court program—the first of its kind in the state

of Minnesota—in July 2010. Since then, there were 527 referrals to the program, with 341 of these

referrals resulting in acceptance to the program. The graduation rate for participants who completed

the program by June 2016 is 75.1%, well above the national average for treatment courts of about

50%.

Finding 1: The program is very successful in connecting participants to Veterans Administration (VA)

services and benefits. Despite a majority of participants entering the program with some level

of connectedness to VA services, almost half of participants saw an increase in benefits due

to participating in Veterans Court. Furthermore, the percentage of participants not receiving

benefits between program entry and exit dropped from 25.2% to 7.7%.

Recommendation 1: Maintain the strong links with the VA representatives on the Steering Committee

and Court Team.

Finding 2: The Veterans Court program is not successful at reducing criminal recidivism when looking

at all participants. Graduates of the program, however, do reoffend at lower rates.

Recommendation 2: Follow the recommendations below that directly apply to graduation rates.

Finding 3: Participants paired with a Veterans Court mentor are significantly more likely to graduate

than participants who have no mentor during program participation. Unfortunately, the

Hennepin County Veteran Court's mentor program was largely dormant throughout the

follow up years of 2014 and 2015.

Recommendation 3: The Veterans Court team should work to expand the newly revamped mentor

program as quickly as possible.

vii


Finding 4: Native American participants have a much lower graduation rate than other racial/ethnic

groups.

Recommendation 4: The program should work to connect Native American participants to more

culturally competent treatment programs and services to reduce this graduation gap.

Finding 5: The program did not rigorously screen participants for illicit substances during its initial

years, despite sobriety being a participant requirement and a central goal of the program.

Recommendation 5: The Veterans Court team should implement a more thorough drug-testing regimen,

with a goal of truly random testing for all participants.

Finding 6: Veterans who enter the program homeless or become homeless during program

participation are more likely to fail the program than participants who have more stable

housing.

Recommendation 6: The Veterans Court team should work to connect homeless participants to housing

resources while adding a housing advocate to the Steering Committee.

Finding 7: Veterans entering the program with more extensive criminal histories are less likely to

complete the program.

Recommendation 7: The Veterans Court team should provide tailor program services and supervision

levels according to criminal history, with additional supervision and services for participants

with more prior convictions.

Finding 8: Participants entering the program with gross misdemeanor charges are much more likely to

graduate, while participants charged with misdemeanor level offenses are more prone to

terminate.

Recommendation 8: The Hennepin County Veterans Court program should adopt a validated risk/needs

assessment to determine which misdemeanor offenders are appropriate for the program and

to tailor services to participants’ documented needs.

1


INTRODUCTION

Since the inception of Drug Courts in the early 1990s, specialized treatment courts focusing on

providing treatment and a heightened level of judicial review for program participants have expanded

in number and scope.1 At their most basic, the overarching aim of specialty courts is to simultaneously

address and rectify the legal issues as well as chemical and mental health needs of participants in these

programs. Through up-front investments in the participants combined with participant commitment

to incorporate positive behavioral changes, these programs hope to facilitate long-term improvements

for participants through reduced subsequent contact with the criminal justice system and enhanced

mental and chemical health outcomes.

One of the most recent innovations in the sphere of specialty treatment courts began in 2008 in

Buffalo, NY, with a program designed to help veterans of the United States Armed Forces in the

criminal justice system. Often termed Veterans Treatment Courts, these programs quickly proliferated

across the United States, numbering 220 by June 2014.2 Veterans Treatment Courts follow the model

of Drug Courts by providing specialized services to meet the individual needs of program participants

in an environment that emphasizes United States military values and culture.

Hennepin County District Court initiated its Veterans Court program—the first of its kind in the state

of Minnesota—in July 2010. This therapeutic jurisprudence model gives veterans of the US Armed

Forces the opportunity to participate in a rigorous program of regularly scheduled judicial reviews and

additional requirements—such as treatment and/or random drug and alcohol testing—with the

incentive of enhancing life stability and potentially receiving a reduced criminal charge or sentence.

The Two Year Review of the Hennepin County Veterans Court Program—which was not a full-scale program

evaluation—generally found that the program was on track to meet its goals,3 however, at that time,

there were an insufficient number of individuals who had exited the program to perform a thorough

evaluation. By June 2016, 155 individuals had exited the program at least two years prior, providing

1 See: http://www.nadcp.org/learn/what-are-drug-courts/drug-court-history 2 See: http://www.justiceforvets.org/vtc-history 3 See: http://www.mncourts.gov/mncourtsgov/media/assets/documents/4/reports/Veterans_Court_Two_Year_Review.pdf

2


an opportune sample with which to evaluate whether the program was meeting its stated goals using

rigorous methodology.

The following document provides a full-scale program evaluation of the Hennepin County District

Court Veterans Court program. It begins with an overview of the program, which describes the

program’s goals and mission, the process of referral and acceptance, and explains how participants

proceed through the program. Next, a robust overview of the 155 participants in the evaluation sample

is presented, which examines participant demographics, criminal history, and military service

background. After a description of the participants, this study analyzes whether the outcomes of

program participants are congruent with the stated goals of the Hennepin County Veterans Court

program. Furthermore, to gain a deeper understanding of the effectiveness of the program, this study

contrasts the outcomes of Veterans Court participants with a matched comparison group of US

Armed Forces veterans who did not participate in the program, which represents an important

innovation of Veterans Treatment Court evaluations. This evaluation concludes with a series of

recommendations and proposed policy refinements to help the program better meet its goals going

forward.

OVERVIEW OF THE HENNEPIN COUNTY DISTRICT COURT VETERANS COURT PROGRAM

The mission of the Hennepin County District Court Veterans Court Program is:

…to promote public safety and assist and support veterans and their families by creating a

coordinated response through collaboration with the veterans’ service delivery system,

community based services, and the criminal justice system.

The specific goals of the program are to:

1) Facilitate participant sobriety

2) Increase compliance with treatment and other court ordered conditions

3) Improve access to VA benefits and services

4) Improve family relationship and social support connections

5) Improve life stability

6) Reduce criminal recidivism

3


The Hennepin County Veterans Court is a voluntary pre-adjudication and post-sentencing program

that utilizes a multi-faceted approach to court supervision. It consists of intensive supervision by a

dedicated probation officer, referral and case management services provided by the Veterans

Administration Medical Center, periodic appearances before the Veterans Court judge, mandatory

chemical and/or mental health treatment support (when applicable), and abstention from alcohol and

illicit drugs (backed by random drug and alcohol testing).

The Hennepin County Veterans Court uses a series of specific guidelines to identify and admit

defendants into the program. While defendants need not be eligible for Veterans Administration

benefits to qualify for the program, potential participants must:

• Be currently serving or have served in the United States Armed Forces

• Be a resident of Hennepin County, MN (although the team can waive the residence

requirement)

• Be at least 18 years of age

• Be charged with a criminal offense in Hennepin County

• Consent to participate in the Veterans Court program

• Not dispute the factual or legal basis of the criminal charges

• Have the prosecutor and defense attorney consent to transferring the case to Veterans Court

• Have a treatable behavioral/mental health and/or substance abuse issue that the defendant is

willing to address via a treatment plan recommended by the Veterans Court team

Regarding prior military service and Veterans Court eligibility across the US, “[Veterans Treatment

Courts] appear to vary in their definition of ‘veteran,’ often depending on discharge status, which is

related to VA eligibility and not actual need for services” (Baldwin 2016: 730, emphasis in original).

One important aspect of the eligibility criteria for the Hennepin County Veterans Court program is

that individuals referred to the program need not be eligible for VA services. To ensure that the

program serves the widest number of individuals with a documented need for services, the program

uses a very broad definition of “veteran” for eligibility, whereby potential participants simply need to

have at least some level of documental military service.

4


Certain factors preclude defendants from participation in the program. Specifically, defendants are not

eligible for the Hennepin County Veterans Court if:

• They face a mandatory or presumptive prison sentence for the alleged offense, according to

the Minnesota Sentencing Guidelines Commission

• They have been identified as a Level 3 Predatory Offender as determined by the Minnesota

Department of Corrections, or they are charged with a felony level criminal sexual conduct

offense

• They are on supervised or conditional release following a prison commitment

Although the program is not employing a validated risk/needs assessment tool to ensure the program

is referring high-risk/high-need individuals (discussed in the recommendation section below), the fact

that the program potentially accepts all but the most egregious criminal offenders helps the program

refrain from cherry-picking defendants who are most likely to succeed.4 By not restricting eligibility

to the program by military discharge type and by allowing the vast majority of criminal offenders

regardless of offense level, the Hennepin County Veterans Court program is actively pursuing

participants who may be more prone to program failure and extending services to defendants with

fewer options.

Yet, meeting the eligibility requirements does not guarantee a defendant referral to the program. For

eligible defendants, referral to the program requires the agreement of the prosecuting and defense

attorneys, the defendant, and the referring judge. After referral to the program, a Veterans Court

screener from the Hennepin County Department of Community Corrections and Rehabilitation

(DOCCR) completes an assessment of the defendant that culminates with a written report to the

Veterans Court team summarizing the findings and highlighting the defendant’s criminal, personal,

and military history.5 A specialist from the Veterans Justice Outreach (VJO)6 also assesses the

4 In contrast, the Veterans Court evaluated by Hartley and Baldwin (2016) restricted the program to defendants facing misdemeanor charges. 5 The DOCCR assessment consists of a long form interview with the defendant, verification of military service history and VA benefit status, a check of the defendant’s criminal history and previous/current utilization of Hennepin County services, and a determination of whether the defendant meets the Veterans Court eligibility criteria. 6 The Veterans Justice Outreach program was created in 2009 “to avoid unnecessary criminalization of mental illness and extended incarceration among Veterans by ensuring that eligible Veterans in contact with the criminal justice system

5


defendant and presents the findings to the Veterans Court team during pre-court staffing (discussed

in more depth below).7

With this information in hand, the Veterans Court team discusses whether to admit the defendant to

the program. When the team chooses not to accept the defendant, the case goes back to the court

calendar in which it began. If the team accepts the defendant into the program, the next steps depend

on the type of charged offense.

For defendants charged with felony level offenses that are not a drug or property crime, the case goes

back to the original judge for resolution and sentencing with the information that the Veterans Court

team will accept the defendant if sentenced into the program. After sentencing, the original judge can

send the case back to Veterans Court, which begins the participant’s tenure in the program in the

Supervised Phase (explained below). Defendants accepted to the program and charged with a

misdemeanor, gross misdemeanor, or drug/property felony are placed on Conditional Release to

Veterans Court (see below). Assignment of a Veterans Court probation Officer occurs upon

acceptance to the program and after the defendant signs the Participant Agreement.8

Once officially in the program, the participant meets with the dedicated Veterans Service Officer to

determine eligibility for additional veterans’ benefits through the Veterans Administration. The team

refers participants eligible for VA services to the VA Medical Center, with the expectation they attend

all appointments and services. For participants who are ineligible for VA benefits, the assigned

probation officer refers participants to community-based services.

One of the main facets of participation in the Veterans Court program is appearing for a series of

regular judicial review hearings to monitor compliance with program requirements. Typically, the

have access to Veterans Health Administration mental health and substance abuse services” (Department of Veterans Affairs 2009). 7 The VJO assessment consists of an interview with the defendant, a check of the VA system to determine current or historical service utilization at the VA Medical Center, and a check of the defendant’s history of compliance with VA care and programming. 8 Additionally, individuals currently on supervision with DOCCR after a convicted offense are eligible for referral to Veterans Court in the event of a probation violation or a post-disposition offense. These probation referrals go through the screening and team decision process outlined above. If accepted, a Veterans Court probation Officer will assume supervision of the participant with the appropriate probation terms and conditions modified accordingly.

6


judge will spend several minutes hearing updates from the participant and the participant’s probation

officer during these judicial reviews. While the judge can reward positive behavior with incentives, the

judge also has the authority to sanction participant behavior that runs counter to court requirements.

Specifically, once in the Hennepin County Veterans Court program, all participants must:

• Remain law abiding9

• Abstain from alcohol and illegal drugs

• Comply with the tenets of their individual treatment plan10

• Take all medication as prescribed

• Maintain contact with probation

• Submit to all drug and alcohol testing as ordered by the Court and directed by probation

• Cooperate with unscheduled home visits by probation and/or law enforcement

• Pay fines, fees, and restitution as ordered by the Court

• Reside in the metropolitan area with accessibility to the Court, probation, and VA services

• Inform the court of address changes

• Complete sentence-to-service or community service as ordered

• Attend self-help or community support groups11 as ordered

• Participate in training and education as ordered

• Seek stable and sober housing

• Adhere to a curfew as ordered

• Be respectful to the Veterans Court team and other Veterans Court participants at all times

9 If a participant picks up a new criminal charge while active in Veterans Court and is unable to resolve the new case with the Veterans Court team, the new case returns to the appropriate criminal court calendar for disposition and sentencing while Veterans Court continues processing the original case continues. If the resolution of the new case includes a probation sentence, the participant’s assigned Veterans Court probation Officer will supervise probation. When the new charge meets the disqualification criteria listed above, the original case(s) leaves Veterans Court until the new case is resolved. If the disposition of the new case does not violate the eligibility criteria mentioned above, the cases can return to Veterans Court after resolution. 10 The VA, community-based treatment providers, probation, and/or the Veterans Court team can establish components of the treatment plan, which could include mental health, substance abuse, domestic violence, or anger management treatment/programming 11 E.g. AA, NA, CA, SMART Recovery, etc.

7


While failure to comply with these provisions can lead to sanctions, participants adhering to these

requirements and making satisfactory progress advance through the program’s phases towards

graduation.

Court Phases

The Hennepin County Veterans Court Program has three basic phases: the Conditional Release

Phase,12 the Supervised Phase, and the Administrative Phase.13 As noted below, not all participants

follow the same phase progression through the program.

Conditional Release Phase

Defendants who enter the Hennepin County Veterans Court program before their cases are resolved

begin their tenure in the Conditional Release Phase. Defendants entering the program in the

Conditional Release Phase can opt out anytime during their first three Veterans Court judicial reviews

or 90 days from their first appearance.14 Once participants have appeared three times before the

Veterans Court Judge or have exceeded the 90-day interval, they can no longer opt out of the program.

During this window, participants are responsible for complying with all of the conditions and

requirements of Veterans Court, and the Veterans Court Judge can sanction these participants for

non-compliance.

All told, the Conditional Release Phase can last up to six months. If a participant’s criminal case is not

resolved within six months of entry to the program, the case returns to the appropriate criminal court

calendar for resolution,15 which allows advancement to the Supervised Phase.

12 Some participants will not participate in the Conditional Release phase due to the severity of the charge that brought them into Veterans Court or if their case is resolved prior to entering the program. These participants will effectively begin their participation in the Supervised Phase. 13 While these phases represent how participants proceed through the program at the time of writing, the Veterans Court phases were not explicitly codified during the era in which the participants in this evaluation were active in Veterans Court, thus there is no information on phase advancement in this evaluation. 14 This opt-out clause exists because a judge has not formally sentenced a participant to the Veterans Court program at this point. 15 Participants remain active in the program while their cases are sent back to a non-Veterans Court calendar for resolution.

8


Supervised Phase

Typically, once a judge sentences a participant to Veterans Court, the participant begins the Supervised

Phase of the program. At this point, the participant can no longer opt out of Veterans Court. In

accordance with court requirements, participants must continue to meet regularly with their probation

officer and attend all scheduled judicial reviews. Furthermore, participants eligible for VA benefits

must attend all screenings, groups, treatment, and appointments as recommended by Veterans

Administration Medical Center staff.16 Participants who are not eligible to receive VA benefits will

work with probation to assess treatment needs and, if necessary, attend treatment, therapy, and/or

aftercare. Judicial reviews tend to become less frequent for participants proceeding satisfactorily

through the Supervised Phase, and participants meeting certain milestones can graduate from this

phase and advance to the Administrative Phase.

Administrative Phase

After meeting the graduation requirements specified below, participants remain on administrative

probation with their Veterans Court probation officer. Although no longer subject to judicial reviews,

participants in the Administrative Phase are required to remain law abiding and must adhere to any

additional requirements deemed appropriate by the Veterans Court judge (e.g. continued abstinence

from alcohol and illicit drugs). The Administrative Phase continues until probation expires or the

Veterans Court judge orders early discharge from probation. Participants who continue to

demonstrate positive behavioral changes developed during Veterans Court and who have not picked

up new criminal charges can request early discharge from probation, thereby fully ending their tenure

in Veterans Court.

Graduation Requirements

In order to move from the Supervised to the Administrative Phase—which effectively allows

participants to graduate from the Hennepin County Veterans Court program—participants must

satisfy several general requirements. First, participants must have resolved all pending criminal court

cases. In addition, participants must not have garnered any new criminal charges for at least six months

16 The Veterans Justice Outreach Specialist will work with participants to facilitate screenings and appointments.

9


and must have demonstrated sobriety for at least six months. Finally, on an individual basis,

participants must have adhered to all specific Veterans Court conditions ordered by the judge. Upon

graduation from the Supervised Phase, participants receive a graduation certificate and a

commemorative coin, and have the opportunity to address the courtroom to reflect on their tenure in

Veterans Court and the positive changes they have made since beginning the program.

Termination Criteria

Participants who fail to comply with program requirements are subject to termination from the

program. Ahead of program termination, the Veterans Court team will work with the participant to

improve motivation and performance. The Veterans Court team rarely terminates participants before

exhausting all avenues to enhance their success. In addition, when participants are absent from the

program for a year, the Veterans Court team discusses the case and the Veterans Court judge ultimately

decides whether to terminate the participant.17 While participants cannot opt out of Veterans Court

once they reach the Supervised Phase, they can request execution of their sentence, which effectively

ends their tenure in Veterans Court.

Program Re-engagement

Participants who graduate or terminate from the program are eligible for subsequent participation in

Veterans Court. Graduates who violate the terms of their probation or pick up a new criminal charge

can have the case referred to the Veterans Court team via the process described above and—if

accepted—begin a new tenure in the program. Participants previously terminated for non-compliance

or prolonged absence from the program can also have their case(s) re-referred to the program pending

approval from the team.18

17 Defendants terminated from the program for being absent can request reinstatement to the Veterans Court program once they return to Hennepin County District Court. The Veterans Court judge, with input from the team, decides whether to reinstate the defendant and impose appropriate sanctions (which can include execution of the participant’s entire sentence). 18 Participants previously terminated and re-referred to Veterans Court likely have to justify what changes they have made in their lives that will enhance the prospects for success.

10


THE VETERANS COURT PROGRAM IN ACTION: REVIEW HEARINGS AND THE VETERANS COURT TEAMS

Court sessions for the Hennepin County Veterans Court program take place every week. Except under

extraordinary circumstances, participants only appear in court at regularly scheduled intervals. The

frequency of these court appearances largely depends upon how long the participant has been in the

program and how well the participant is faring in the court. The goal is that review hearings become

less frequent the longer a participant is active in Veterans Court and complies with program

requirements. When first accepted, participants are generally required to attend review hearings every

30 days. Participants making satisfactory program progress who are nearing graduation might have as

long as 90 days between scheduled judicial reviews. Participants whose compliance with program

requirements is lacking are subject to more frequent review hearings, potentially having to appear in

court weekly. Team members, particularly the participant’s probation officer, discuss and agree on the

frequency of the judicial reviews.

The creation of policies guiding the day-to-day operations of the programs and the carrying out of

these policies are the responsibilities of two different—but often overlapping—teams: The Steering

Committee and the Court Team.

Steering Committee

The Veterans Court Steering Committee is a group of high-level representatives from the different

departments and bureaus that have a stake in the program.19 This group meets monthly with the

general mission of monitoring adherence to the program’s established policies and procedures, as well

as compliance with the standards and best practices guiding Veterans Treatment Courts. The Steering

Committee discusses, decides, and implements all potential policy changes.

19 While all interested parties may attend and participate in Steering Committee meetings, the voting members of the committee are the presiding judge, the coordinator, the probation supervisor, the VA Justice Outreach specialist, the Hennepin County Veterans Service Officer, VA Benefits representative, the mentor coordinator, the Minneapolis City Attorney, the Hennepin County Attorney, the Hennepin County Public Defender, the Minneapolis Police Department representative, the Minnesota Department of Education and Employment Development (DEED) representative, and the Hennepin County District Court research analyst.

11


Court Team

Whereas the Steering Committee sets the policies and procedures the for Veterans Court program,

the Court Team is responsible for carrying out these policies in the courtroom. The Court Team meets

before the weekly court session for staffing, where they discuss the progress of the participants

appearing in front of the judge that day. In addition, the Court Team decides whether to accept

defendants into the Veterans Court program and determines when participants are ready to graduate.

While members of the Court Team can recommend sanctions or termination for participants who do

not comply with program requirements, the Veterans Court judge is solely responsible for imposing

sanctions for non-compliance, up to and including termination from the program.

Six Years of the Hennepin County Veterans Court Program

The program’s policies, procedures, and program phases have evolved throughout the six years the

court has been in operation. Over this time span, there have been three different presiding judges.

Furthermore, after about two years of operation, a formal Veterans Court Coordinator position was

established.

From July 2010 to June 2016, there were 526 referrals to the Hennepin County Veterans Court

program.20 Nearly two-thirds (64.8% or 341) of these referrals resulted in acceptance to the program,

while 186 of these referrals did not. Of the 341 admittances to the program, 60.9% (208) resulted in a

successful graduation from the program, 20.2% (69) in termination, 7 in-program deaths, with the

remaining 16.4% (56) as active participants. Looking exclusively at those who exited the program, the

graduation rate between July 2010 and June 2016 was 75.1%, well above the national average Drug

Court graduation rate of about 50% (Mitchell et al. 2012).

The acceptance rate for the program during this period was 64.1%. While the rationale for non-

acceptance were not documented in all instances, the most common reason was because the defendant

was not interested in the program and/or refused to participate. The second most common reason

for non-acceptance was for defendants failing to appear at the required screening. Thus, defendants

20 These referrals do not represent unique individuals. Several individuals had multiple referrals to the program.

12


themselves were typically the main reason for non-acceptance to the program; indeed, in only a

handful of documented cases did team members object to accepting a referral.

RESEARCH DESIGN

At its most basic, the purpose of this evaluation is to determine whether the program is meeting its

goals of facilitating participant sobriety, increasing participant compliance with treatment and other

court-ordered conditions, improving participant access to VA benefits and services, improving family

relationships and social support connections, improving life stability, and reducing criminal recidivism.

The Sample

This evaluation examines the outcomes for 155 program participants who participated in Veterans

Court between 2010 and 2013 and exited the program at least two years before 2016. Although best

practices suggest examining a three-year post-program window to assess program effectiveness,21 this

was not feasible for the present study. Given that Veterans Court began in Hennepin County in 2010

and that participants spend an average of about twelve months in the program, there would have been

too few individuals with three years of “street time”22 to draw reliable results. Nevertheless, two years

of follow-up represents an adequate window of time to demonstrate whether the program is meeting

its goals, especially since most of the program goals pertain to the time when participants are active in

the program.

By comparing Veterans Court participants to a matched comparison sample of military veterans who

did not participate in the program identified through the propensity score matching technique, this

study overcomes a significant limitation of the Two Year Review of the Hennepin County Veterans Court

Program.

While propensity score matching still falls short of the “gold standard” of random assignment,

including a statistically matched comparison group of veterans represents an important innovation of

this study. Indeed, Veterans Court evaluations typically lack a matched comparison group of veterans

21 See: http://www.nadcp.org/Standards 22 The term “street time” refers to the amount of time that offenders were not in jail, since incarcerated individuals are much less capable of committing new crimes.

13


who were not Veterans Court participants. Furthermore, when evaluations have made use of matched

comparison groups, they have compared program participants against individuals who chose not to

participate after a referral to the program (e.g. Hartley and Baldwin 2016), which does not accord with

NADCP’s standards for equivalent and unbiased comparison groups (NADCP 2015).

Methods

This study evaluates the effectiveness of the Hennepin County Veterans Court program using a variety

of methods. At the most basic level, this study shows the longitudinal arc of program participants

from acceptance to exit, contrasting participants with themselves before and after Veterans Court.

Specifically of interest is the degree to which Veterans Court participants improved certain life quality

metrics, such as employment status, housing status, educational attainment, and prosocial

activities/relationships, in accordance with the program’s stated goals.

In determining whether the program met its goal of reducing criminal recidivism, this study examines

whether program participants had fewer post-Veterans Court convictions than the comparison group.

As explained below, the comparison sample was statistically similar to the Veterans Court cohort.

Moreover, no comparison group members had prior referrals to any of Minnesota’s treatment courts

(such as Drug Court, DWI Court, or Mental Health Court).

Having a comparison group that is statistically identical to the Veterans Court cohort will allow us to

determine whether there was a beneficial “program effect” of the Hennepin County Veterans Court

program. Also analyzed is how quickly and how often the respective samples reoffend.

This study also analyzes the factors that predict program success or failure via regression analysis.

Understanding the program features that enhance participants’ prospects for success as well as the

participant attributes associated with higher rates of failure can help the Hennepin County Veterans

Court tailor its policies, procedures, and services to give every participant the upmost opportunity to

graduate.

14


Data Sources

Data for this evaluation come from a variety of sources:

• Veterans Court Policy and Procedure Manual. This handbook explains the processes of referral and

acceptance to the program, as well as the basic outlines of how the program works. As the

guiding document for how the program is meant to function in practice, the Manual helps

discern the degree to which court policies and procedures are followed and whether these

policies need fine tuning in response to the outcomes in this study

• MNCIS (Minnesota Court Information System). Demographic and criminal case information for

the Veterans Court cohort and the comparison group comes from Minnesota’s Court-

management Information System. This statewide database was also used to compile

information on criminal history and recidivism for the Veterans Court cohort and the

comparison group as well as an array of demographic factors

• Veterans Court Screening Documents. As mentioned above, a member of the Hennepin County

DOCCR screens individuals referred to the Veterans Court program. These data typically

include information about branch of military service, rank, deployment history, whether or

not they saw combat, and military discharge type

• Hennepin County District Court Treatment Court Database. In late 2014, the Hennepin County

Research Department developed a database containing information for former and current

participants in all four Treatment Courts (Veterans, Criminal Mental Health, DWI, and

Drug).23 Information for prior Veterans Court participants was obtained from multiple

sources:

o Veterans Court database maintained by DOCCR. Through 2013, the Hennepin County

DOCCR collected data on Veterans Court participants at entry and exit. The relevant

data points were transferred from DOCCR’s database to the Research Department

database

o Probation data questionnaires. In order to get participant information that was not in the

DOCCR Veterans Court database, DOCCR staff undertook an extensive examination

23 A dedicated employee keeps this database up to date, which will facilitate the expedient completion of future Treatment Court evaluations.

15


of Veterans Court participant records and completed paper forms containing

necessary intake and exit data during the summer of 2015.

o Court Services Tracking System (CSTS). DOCCR stores information about individuals on

probation in their information system named CSTS. While this system primarily

provided information about drug test results, the detailed client case notes written by

probation officers helped corroborate other participant data.

Limitations

One large limitation of this study is that the VA was unable to provide military service information

and VA service utilization for the comparison sample (explained in more detail below). Veterans Court

participants sign a consent form that allows data sharing between the VA and justice system partners

for a specified amount of time, which allows the Veterans Court Team to access a host of data about

program participants. Because the individuals in the comparison group did not consent to any type of

data sharing, their protected information remains private. These data privacy rules even precluded the

VA from providing aggregate data about the comparison group without any individual identifying

information. In addition, no data on housing status, employment, or education history were available

for the comparison group.

Unfortunately, since evaluations are retrospective in nature, some information is missing for some

participants despite the best efforts of the DOCCR Veterans Court team to collect as much historic

data as possible.

In addition, the calculation of “street time” for the recidivism window (explained in more detail below)

relied on certain assumptions. Specifically, while the recidivism window was two years, the Research

Division took pains to account for the number of days individuals spent in jail during this window

and were unable to recidivate. The two-year street time window considers the number of days an

individual spent in jail to ensure that everyone had a full 730 days of non-incarcerated street time

during which to reoffend. One limitation is that MNCIS does not capture jail days for individuals who

might have spent time in jail for non-convicted offenses; that is, incarceration time is only available

for individuals arrested, booked, and subsequently convicted of the charge. However, this limitation

16


affects both the Veterans Court cohort and the comparison group in similar ways, and there is no

reason to suspect that one group would have more or less non-recorded jail time than the other group.

Despite these limitations, the addition of a matched comparison groups—even without military

service information, VA treatment utilization, or life stability metrics—allows us to have a high degree

of confidence in the results presented below. Furthermore, the fact that “there is currently a dearth of

research on [Veterans Courts] in general” ensures that the this study—by virtue of its rigorous

methodology—is an important addition to the extant Veterans Court literature (Baldwin 2016: 735).

DEMOGRAPHIC PROFILE OF VETERANS COURT COHORT

Before analyzing whether the program is or is not meeting its goals, this study examines the profiles

of the 155 Veterans Court participants in the evaluation sample in detail via side-by-side comparisons

of graduates and non-completers.

Gender

Overall, the Veterans Court participants in this study are overwhelmingly male, as Table 1 below

demonstrates. Specifically 150 (96.8%) of the participants in the evaluation sample are male, while five

(3.2%) are female. While the population of the US military as a whole is predominantly male, 16.5%

of Active Duty and Selected Reserve members are female (Department of Defense 2014: 6), which is

five times greater than the proportion of females in the Veterans Court program. However, given that

the criminal defendant population also skews heavily in favor of men, the large gender disparity in the

Hennepin County Veterans Court program is hardly surprising. The paucity of female participants

renders it difficult to discern whether gender affects program outcomes.

Table 1: Gender Breakdown of Veterans Court Participants (N=155)

Non-Completers Graduates Total (% of Total)

Female 2 3 5 (3.2%) Male 34 116 150 (96.8%)

17


Race

Table 2 below documents the self-reported race/ethnicity of the Veterans Court participants in this

study. The largest racial categories in Veterans Court are White (65.8%) and African American

(26.5%), which is unsurprising given that these groups comprise the two largest racial groups in the

US Armed Forces, (71.0% and 16.8%, respectively) (Department of Defense 2014: 7).

Table 2: Race/Ethnicity Breakdown of Veterans Court Participants (N=155)


Native American or Alaskan Native 5 1 6 (3.9%)

Asian 0 2 2 (1.3%)

Black or African American 11 30 41 (26.5%)

Hispanic/Latino 1 3 4 (2.6%)

White 19 83 102 (65.8%)

Figure 1 below looks at the program success rate (the proportion of participants from each

racial/ethnic group who completed the program successfully) of the various racial and ethnic groups.

This figure displays the major difference in success rates for Native Americans. Whereas all other

racial and ethnic groups had a success rate in the program of at least 73%, the success rate for Native

Americans was only 16.7%. Although the number of Native American participants in the sample is

quite small, there is a meaningful difference between Native Americans and all other racial/ethnic

groups by program outcome, whereby Native Americans are more likely to end up terminating from

the program. Given the high termination rates of Native American participants—despite the small

sample size—one recommendation of this study is to examine ways to integrate culturally competent

programming or other services to help reduce the graduation gap for Native American participants.

18


Figure 1: Veterans Court Success Rates, by Racial/Ethnic Group

Age

Table 3 below contains descriptive statistics for the age of participants at the time of program

acceptance. The average age for all participants when they came into the program was 43.4 years. The

youngest participant was 20 and the oldest was 66. While non-completers tended to be slightly older

at the time of acceptance into Veterans Court than program graduates were, the difference was not

statistically significant.

Table 3: Age of Veterans Court Participants, Descriptive Statistics (N=155)

Mean Minimum Maximum Std. Dev.

Graduates 43.0 20 66 13.5

Non-Completers 44.6 22 63 12.9

All Participants 43.4 20 66 13.3

100.0%(N=2)

81.4%(N=102) 75.0%

(N=4)73.2%

(N=41)

16.7%(N=6)

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

80.0%

90.0%

100.0%

Asian White Latino Black Native American

19


Employment

Table 4 examines the employment status of participants at the beginning of their tenure in Veterans

Court. Overall, participants were somewhat bifurcated in their employment status at program entry,

whereby a plurality of participants were unemployed (29.0%) followed closely by participants who

were employed on a full-time basis (27.1%).

It is worth noting that individuals who began the program working full- or part-time tended to

graduate at very high rates, suggesting that these participants might be primed for success and may

need fewer services or attention than individuals who have less stable employment statuses at entry.

Table 4: Employment Status at Veterans Court Entry (N=155)


Unemployed 16 29 45 (29.0%)

Disabled 9 21 30 (19.4%)

Retired 0 6 6 (3.9%)

Student 2 7 9 (5.8%)

Part-time 1 17 18 (11.6%)

Full-time 6 36 42 (27.1%)

Unknown 2 3 2 (3.2%)

Education

As Table 5 indicates, it was most common for Veterans Court participants to have taken some college

courses (but did not graduate) when they began the program (37.4%), followed by about 30% of

participants who were high school graduates/GED recipients. Although the vast majority of

individuals who entered the program with at least a two-year degree completed the program

successfully, so did 80% of individuals who had not graduated high school (albeit with a much smaller

sample size). Similar to employment status, participants who enter the program with at least a 2-year

degree graduate at high rates, suggesting that it could be appropriate to divert certain program

resources from participants with a higher level of education.

20


Table 5: Education Status at Veterans Court Entry (N=155)


Less than High School 2 4 6 (3.9%)

High School Diploma/GED 17 29 46 (29.7%)

Some College 12 46 58 (37.4%)

Two-Year Degree 2 23 25 (16.1%)

Four-Year Degree and Up 3 17 20 (12.9%)

Housing Status

A majority of Veterans Court participants had independent housing when they began the program

(56.8%), as Table 6 displays. Living with a relative or friend was also quite common at the time of

Veterans Court entry (26.5%). Program success was highly correlated with individuals in both of these

housing statuses. On the face of things, individuals who were homeless at the outset of Veterans Court

are more likely to terminate from the program than graduate (corroborated by regression analysis

below). While one program goal is increasing housing stability, homeless veterans appear to face

unique challenges that imperil their success in the program. As a result, these participants may require

additional or different services than those who enter the program with stable housing.

Table 6: Housing Status at Veterans Court Entry (N=155)


Homeless 10 5 15 (9.7%)

Residential Facility 4 7 11 (7.1%)

Relative/Friend 7 34 41 (26.5%)

Independent 15 73 88 (56.8%)

21


Table 7: Charge Level at Disposition for Veterans Court Instant Offense (N=155)


Misdemeanor 19 35 55 (35.4%)

Gross Misdemeanor 6 64 70 (45.2%)

Felony 11 109 30 (19.4%)

Offense Type

Overall, it was most common for participants to enter Veterans Court facing gross misdemeanor

charges (45.2%), as Table 7 above displays. Misdemeanor charges were slightly less common (35.4%),

while the number of participants facing felony level charges lagged behind (19.4%). As Figure 2 shows,

the most common specific offense types for Veterans Court participants were Gross Misdemeanor

DWIs (35%) followed by Other Felonies (19%), Other Non-Traffic Offenses (16%), Misdemeanor

Domestic Assaults (15%), Other Gross Misdemeanors (10%), and Misdemeanor DWIs (5%).

Figure 2: Veterans Court Instant Offense by Type

Other Non-Traffic16%

Misdemeanor DWI5%

5th Degree Assault15%

Other Gross Misdemeanor

10%

Gross Misdemeanor DWI35%

Other Felony19%

22


As analyzed below in more detail, misdemeanant participants are more likely to terminate than those

facing gross misdemeanors or felonies. In contrast, gross misdemeanant participants are significantly

more likely to succeed than those facing misdemeanor or felony charges. The diverging success rates

of participants with different instant offenses offers some important policy implications explored in

more depth in the recommendation section below.

Prior Convictions

Table 8 provides descriptive statistics for the criminal history of the Veterans Court participants at

program entry. The top three-fourths of Table 8 display the average number of prior misdemeanor,

gross misdemeanor, and felony convictions for Veterans Court participants, broken down by

graduates and non-completers. Across all these offense levels, there is a statistically significant

difference by program outcomes, whereby individuals who wound up not completing the program

had a higher number of average prior convictions at program entry than eventual graduates.

Another way to conceptualize criminal history is through a point system that accounts for the number

and severity of prior convictions. The bottom quarter of Table 8 examines the average number of

criminal history points for graduates and non-completers. The point system gives four points for a

“person felony” offense (e.g. a felony domestic assault), three points for a “non-person felony”

offense (e.g. a felony drug offense), two points for a “non-felony person” offense (e.g. misdemeanor

assault), and one point for “non-felony non-person” offenses (e.g. misdemeanor DWI), multiplied the

number of convictions at each offense level. Thus, if an individual had one prior felony person

conviction (four points) and two non-felony person convictions (two points each), the criminal history

point total would be eight. Not surprisingly, this method of measuring prior criminal activity shows

that eventual non-completers enter the program with significantly more average criminal history

points than eventual graduates (10.0 and 2.8 points, respectively). Given this graduation gap based on

criminal history, the Veterans Court team should provide a different array of services and/or higher

levels of supervision for individuals entering the program with more extensive criminal histories. In

contrast, individuals with no criminal convictions prior to entering the program likely require less

supervision and fewer services.

23


Table 8: Convictions before Entering Veterans Court

Prior Misdemeanor Convictions Mean Minimum Maximum Std. Dev. Graduates 1.2* 0 9 1.6

Non-Completers 4.4* 0 35 7.2


Prior Gross Misdemeanor Convictions Mean Minimum Maximum Std. Dev. Graduates 0.6* 0 6 1.1 Non-Completers 1.3* 0 11 2.4


Prior Felony Convictions Mean Minimum Maximum Std. Dev.

Graduates 0.3* 0 8 1.0



Criminal History Points Mean Minimum Maximum Std. Dev. Graduates 2.8* 0 24 4.5 Non-Completers 10.0* 0 56 14.4

All Participants 4.5 0 56 8.5 * Difference is statistically significant at the .001 level

Program Participation Statistics

Tables 9 and 10 below examine program participation data for Veterans Court participants. Table 9

looks at the average number of months spent in the program. Not surprisingly, graduates are in the

program significantly longer than participants who do not graduate. Program graduates spend a little

over 14 months in the program on average while non-completers average just over six months in the

program, a statistically significant difference.

24


Table 9: Months Spent in Veterans Court Program


Graduates 14.2* 5 34 5.8


All Participants 12.3 0 34 6.7 * Difference is statistically significant at the .001 level

Despite the fact that graduates spend a longer time in the program, they do not appear in front of the

Veterans Court judge significantly more than non-completers. As Table 10 demonstrates, program

graduates average about 14 review hearings during their time in the program—an average of about

once a month during their tenure. Non-completers average over 12 hearings for the duration of their

participation, which equates to about two hearings per month, on average, for these individuals.

Table 10: Number of Veterans Court Review Hearings


Graduates 14.0 5 46 7.0



It is unsurprising that individuals who eventually terminate are subject to a higher level of judicial

supervision as they likely demonstrate behavior that does not earn the trust of the Veterans Court

team. However, requiring less compliant participants to attend a greater number of review hearings

does not appear to enhance program success on its own. As a result, the Hennepin County Veterans

Court program should look at additional ways outside of the courtroom to help participants who are

most at risk of failure forge a successful path.

Military History

The most basic unifying thread of Veterans Court participants is that they all have previously served

in the United States Armed Forces. Figure 3 below enumerates the branch of service for the

25


participants in Veterans Court.24 Active Duty Army members comprised the most common branch

for Veterans Court participants (41%), followed by Active Duty Marine Corps members, and Army

National Guard members. All told, the vast majority of participants were active duty members in their

respective branch as opposed to reservists, suggesting they are likely to identify with the military

structure and values built into the Hennepin County Veterans Court program.

Figure 3: Branches of Military Service of Veterans Court Participants

As Table 11 denotes, the vast majority (88.4%) of Veterans Court participants received an honorable

discharge. Interestingly, the two participants discharged from the Armed Forces for bad conduct—

one of the lowest types of discharges the US military offers—successfully graduated from the program.

Only the discharge category “General Discharge under Honorable Conditions” had more terminated

participants than graduates (although the total number of participants within this discharge category

was quite small). Thus, discharge type does not appear to impact program outcomes.

24 The number of military branch instances adds up to more than 155 because ten participants were in multiple branches during their military careers.

69

2926

1813

5 3 2 1 10

10

20

30

40

50

60

70

Army(activeduty)

MarineCorps(activeduty)

ArmyNationalGuard

Navy(activeduty)

Air Force(activeduty)

ArmyReserve

NavyReserve

MarineCorps

Reserve

Air ForceReserve

AirNationalGuard

26


Table 11: Discharge Type of Veterans Court Participants (N=155)


Bad Conduct Discharge 0 2 2 (1.3%)

Other than Honorable Discharge 2 5 7 (4.5%)

General Discharge under Honorable Conditions 5 4 9 (5.8%)

Honorable Discharge 29 108 137 (88.4%)

Another important facet of participants’ military service history is rank. Table 12 below lists the rank

at discharge for all participants, with the hierarchy of the military ranks going from lowest (E1; the

“E” designates enlistees) to highest (O2; the “O” designates officers).25 As the rightmost column of

Table 13 demonstrates, a plurality of participants exited the military at a rank of E4, while the next

most popular rank at discharge was E1, which is the lowest rank possible. While there were a higher

percentage of non-completers at certain ranks—E1 through E4—most participants within these ranks

successfully completed the program. Interestingly, although they represent only a fraction of total

program participants, all participants above the rank of E5 successfully graduated from the program.

Table 12: Rank at Discharge from Armed Forces (N=155)


E1 10 18 28 (18.1%)

E2 6 6 12 (7.7%)

E3 6 19 6 (3.9%)

E4 11 45 56 (36.1%)

E5 3 19 22 (14.2%)

E7 0 6 6 (3.9%)

E8 0 1 1 (0.6%)

O1 0 1 1 (0.6%)

O2 0 4 4 (2.6%)

25 For more information on military ranks, see: http://www.defense.gov/About-DoD/Insignias

27


One of the rigors of the US Armed forces is deployment abroad. More than half of the participants

in Veterans Court (55.6%) had a deployment abroad at some point during their military service as

Table 13 below indicates.

Table 13: Deployment History of All Veterans Court Participants (N=155)


No 22 47 69 (44.5%) Yes 14 72 86 (55.5%)

A further sacrifice many members of the US military make is deployment in a forward combat theater.

Table 14 below shows the number of participants who saw combat among the 86 deployed

participants (Table 14 excludes non-deployed participants). The bulk of deployed Veterans Court

participants saw combat during their military service (79.1%), and the vast majority of these individuals

graduated from the program (58 out of 68). Thus, it does not appear that individuals who saw combat

were more in jeopardy of terminating from the program. Instead, it is possible that some of the

requirements of being in a combat situation—teamwork, adherence to protocol, and personal

discipline—might prime these individuals to be more successful in the Veterans Court program.

Table 14: Combat Deployment History of Veterans Court Participants (of the Participants Deployed, N=86)


No 4 12 16 (18.6%) Yes 10 58 68 (79.1%)

Unknown 0 2 2 (2.3%)

Total 14 72 86 (100.0%)

In sum, the average participant is white and male, and discharged honorably from Active Duty Service

in the Army. Graduates tend to have less extensive criminal histories and face gross misdemeanor

28


charges. Participants who fail to complete the program are often have more prior convictions and

enter the program on misdemeanor charges.

Examination of the Comparison Group

NADCP’s best practice standards for treatment courts stipulate that “outcomes for [treatment court]

participants are compared to those of an unbiased and equivalent comparison group” (NADCP 2015:

60). Unfortunately, since Hennepin County District Court and its justice partners did not

systematically collect military service history, it appeared, at first, impossible to identify a group of

veterans in the criminal court who did not participate in the program.26 Nevertheless, with the help

of multiple agencies, the Hennepin County District Court Research Division identified a matched

comparison group of US Armed Forces veterans convicted of criminal charges in Hennepin County

District during the same years that the Veterans Court cohort participated in the program (2010

through 2013). Unlike Hartley and Baldwin (2016), the study did not include individuals who had a

referral to the Veterans Court program but chose to opt out as potential comparison group members.

In fact, none of the individuals in the comparison group had a referral to any Treatment Court in the

entire state of Minnesota.

The identification of a comparison sample of justice-involved veterans was a multi-step process. First,

the Hennepin County District Court Research Division categorized the Veterans Court cohort by

their instant offense type, age, gender, race, and criminal history.27 Using MNCIS, the Research

Division compiled profiles for several thousand individuals convicted of a criminal offense in

Hennepin County District Court from 2010 to 2013 and did not have a referral to any treatment courts

throughout Minnesota. By virtue of an interagency data sharing court order and cooperation between

the St. Paul Veterans Benefit Administration (VBA) and the Hennepin County District Court, the

VBA identified 216 Hennepin County criminal offenders with verified military service.

26 This problem is not isolated to Hennepin County. In fact, it is extremely common for jurisdictions not to capture veteran status. For example, Baldwin (2016: 734) notes that 88% of jurisdictions with Veterans Treatment Courts do not have “a set procedure for the identification of veterans in contact with the criminal justice system.” 27 Measured by a tally of criminal history points based upon the number and severity of prior convictions as explained above in the section documenting the criminal history of the Veterans Court cohort.

29


The Hennepin County District Court Research Division used propensity score matching (PSM) to

select individuals from this pool of 216 veterans who were statistically identical to the Veterans Court

cohort based on the aforementioned variables.28

Table 15: Instant Offense Charge Type of Comparison Group and Veterans Court Cohort

Comparison Group Veterans Court Total (% of Total)

Other Non-Traffic 21 (13.6%) 24 (15.5%) 45 (14.5%)

Misdemeanor DWI 8 (52%) 8 (5.2%) 16(5.2%)

5th Degree Assault 23 (14.8%) 23 (14.8%) 46(14.8%)

Other Gross Misdemeanor 14 (9.0%) 15 (9.7%) 29 (9.4%)

Gross Misdemeanor DWI 54 (34.8%) 55 (35.5%) 109(35.2%)

Other Felony 35 (22.6%) 30 (19.4%) 65 (21.0%)

Starting with the charge level for the instant offenses seen in Table 15, the comparison group had a

few less Other Non-Traffic offenses, several more Other Felony offenses, and a reasonably similar number

of all other offense types compared to the Veterans Court cohort. 29 Despite the fact that the charge

levels are not exactly the same between each group, a chi-squared test of the differences between the

28 In order to ensure there was a match for each member of the Veterans Court cohort, the Research Division used the “one-to-many” matching option, whereby certain individuals in the comparison group appear more than once. The Research Division’s statistical software applies weights to individuals in the comparison group who appear multiple times to ensure they match the number of individuals in the treatment group. Specifically, the comparison sample contains 98 unique individuals, but the statistical weighting technique transforms these 98 individuals into 155 observations. While the vast majority of individuals in the comparison group appear only once, 10 individuals are included at least three times, with one individual in the comparison group accounting for six observations. The Research Division considered using the one-to-one matching option whereby each member of the Veterans Court cohort has a unique counterpart in the comparison group. However, the one-to-one option did not yield a unique match for eight members of the Veterans Court cohort. The Research Division decided it was better to have the full cohort of 155 Veterans Court participants with a one-to-many matched comparison group than dropping members of the Veterans Court sample in order to have a comparison group with unique matches. Importantly, the Research Division ensured the results of the recidivism analysis below were the same using the one-to-many option and the one-to-one option. 29 The data in Tables 16-19 reflect attributes of the comparison group when their instant offense case was disposed. For example, in Table 19, age for the Veterans Court cohort is their age at the time of acceptance to the program whereas for the comparison group it represents how old each individual was at the time of the conviction for their instant offense.

30


charge levels of the two groups was not statistically significant, suggesting that there is no discernible

difference in the type of instant offense between the Veterans Court cohort and the comparison group.

Similarly, Table 16 examines the differences in the racial composition between the two groups and

shows that there are some differences between the number of Non-White and White individuals in

each group, whereby the comparison group has several fewer Non-White individuals than the

Veterans Court cohort. However, there was no statistically significant difference in the racial

breakdown between the two groups.30

Table 16: Racial Composition of the Comparison Group and Veterans Court Cohort

Comparison

Veterans Court Total (% of Total)

White 108 (58.7%) 102 (65.8%) 210 (67.7%)

Non-White 47 (41.2%) 53 (34.2%) 100 (32.3%)

Table 17 shows that the gender breakdown of the comparison group almost exactly mirrors the

Veterans Court cohort, producing no statistically significant gender differences between the groups.

Table 17: Gender Breakdown of the Comparison Group and Veterans Court Cohort

Comparison

Veterans Court Total (% of Total)

Female 4 (2.6%) 5 (3.2%) 9 (2.9%)

Male 151 (97.4%) 150 (96.7%) 301 (97.1%)

Table 18 below looks at the criminal history of the Veterans Court cohort and the comparison group.

The top three-fourths of the table compare the average number of prior misdemeanor, gross

misdemeanor, and felony convictions, respectively. Although the comparison group presents with a

30 Whereas Table 2 above listed the specific race or ethnicity of the Veterans Court cohort, the analysis below uses a dichotomous race/ethnicity variable that categorizes participants as White or Non-White.

31


higher average number of convictions across all offense levels, the differences are not statistically

significant. The bottom quarter of the table looks at the average number of criminal history points

used above, which is a sum of points based upon the number and severity of prior convictions. Once

again, members of the comparison group averaged a higher number of criminal history points than

the Veterans Court cohort, but this difference was not statistically significant.

Table 18: Prior Convictions of Comparison Group and Veterans Court

Prior Misdemeanor Convictions Mean Minimum Maximum Std. Dev. Comparison Group 2.3 0 15 3.0

Veterans Court 1.9 0 35 3.9

Both Groups 2.0 0 35 3.5

Prior Gross Misdemeanor Convictions Mean Minimum Maximum Std. Dev. Comparison Group 0.9 0 7 1.6 Veterans Court 0.7 0 11 1.5


Prior Non-Felony Convictions Mean Minimum Maximum Std. Dev.

Comparison Group 0.6 0 9 1.4



Criminal History Points Mean Minimum Maximum Std. Dev. Comparison Group 5.3 0 47 7.8 Veterans Court 4.5 0 56 8.5


The final indicator used to match the Veterans Court cohort with the comparison group was age,

which Table 19 displays. The average age of the two groups revealed no statistically significant

differences although the comparison group is slightly older, on average.

32


Table 19: Average Age of Comparison Group and Veterans Court Participants


Comparison Group 44.7 21 73 14.3


Both Groups 44.1 20 73 14.8

All told, the comparison group is not different from the Veterans Court cohort in any meaningful

way—that is, the differences are not statistically significant—suggesting that this group represents an

ideal set of individuals to determine whether the Veterans Court program met its goals.

EVALUATING OUTCOMES OF VETERANS COURT PARTICIPANTS

This section examines the degree to which the outcomes of Veterans Court participants adhere to or

deviate from the stated goals of the program and represents the most vital part of this study.

Depending on the goal under evaluation, this section relies on different methods and points of

comparison to evaluate program outcomes, as mentioned in the Research Methods section. For some

goals, it is most appropriate to examine only the Veterans Court participants, while other goals require

including the comparison group to ascertain outcomes.

Goal 1: Facilitate Participant Sobriety

One uniform condition of the Veterans Court program is that participants remain sober, enforced

through random drug/alcohol testing of participants. One way to measure whether the program is

facilitating participant sobriety is to examine the results of drug tests administered by the Hennepin

County DOCCR during program participation. Table 20 demonstrates that about an equal proportion

of eventual Veterans Court graduates and non-completers had at least one drug test while active in

the program, about 56% (55.6% of graduates and 56.3% of non-completers). Although an equal share

of eventual non-completers and graduates were drug tested during the program, many participants did

not undergo any drug testing by DOCCR at all. This lack of widespread testing is problematic given

that about two-thirds of all Veterans Court participants had a diagnosis of a Substance Use and

33


Addictive Disorder when they exited the program (see Table 22 below), suggesting that many

participants with substance abuse issues were not screened for drug use.

Table 20: Percentage of Veterans Court Participants Subjected to In-Program Drug Testing by DOCCR (N=155)


At least one test 16 (44.4%) 52 (43.7%) 68 (43.9%)

No drug tests 20 (55.6%) 67 (56.3%) 87 (56.1%)

For the participants who were tested, the average number of drug tests during Veterans Court was

hardly different between graduates and non-completers, with each group averaging just over 10 tests

as seen in the top portion of Table 21. However, non-completers spent eight fewer months on average

participating in the program (see Table 9 above); therefore, DOCCR tested non-completers on a more

regular basis than they tested graduates. Indeed, the middle portion of Table 21 shows that while

graduates averaged about 0.8 drug tests per month while active in the program, non-completers

averaged about 1.6 tests per month, a difference that is statistically significant. Given that non-

completers likely demonstrate non-compliance with program rules before leaving the program, the

increase in drug tests for these individuals is understandable. Nevertheless, the entire cohort of

Veterans Court participants averaged a little less than one test per month. The lack of widespread

testing contravenes the standards and best practices of treatment courts per the NADCP, which

stipulate testing all participants on a truly random basis.31

Unsurprisingly, participants who successfully completed the program averaged significantly fewer

positive tests than non-completers as the bottom portion of Table 21 demonstrates. Interestingly,

testing positive for illicit substances did not automatically imperil program success. In fact, the highest

31 See: http://www.allrise.org/sites/default/files/nadcp/AdultDrugCourtBestPracticeStandards.pdf; pp 38-53.

34


number of positive drug tests for a terminated participant was 16, which is only one more than a

program graduate who failed 15 drug tests.32

Table 21: In-Program Drug Test Information for Veterans Court Participants

Average Drug Tests Administered for Duration of Program Mean Minimum Maximum Std. Dev. Graduates 10.4 0 121 15.5



Average Drug Tests Administered per Month in Program Mean Minimum Maximum Std. Dev. Graduates 0.8** 0 5.5 1.2

Non-Completers 1.6** 0 7 2.2


Average Positive Drug Tests for Duration of Program Mean Minimum Maximum Std. Dev.

Graduates 1.0* 0 15 2.5 Non-Completers 1.9* 0 16 3.9

All Participants 1.2 0 10 1.5 * Difference statistically significant at the .05 level; ** Difference is statistically significant at the .001 level

Unfortunately, it is difficult to follow the trajectory of whether or not Veterans Court participants

maintained sobriety after exiting the program since very few individuals had systematic drug tests after

leaving the program. Specifically, about 31% (11 of 36) of non-completers were drug tested in the two

years following their exit from the program while only 4.2% of graduates (5 out of 119) submitted a

drug test in the two years after graduation—despite the fact that the majority of graduates and non-

completers alike remained on probation supervision of some sort. Looking at drug test results after

32 One reason that positive drug tests might not lead to termination is that probation expects positive test results in certain circumstances. For example, repeated positive drug tests might be able to demonstrate that the presence of a substance—especially THC, which stays in the system longer than other substances—is diminishing, thereby suggesting abstention from this substance over time if levels of the substance continually drop. As a result, a “positive” test might paradoxically demonstrate movement towards sobriety, although these circumstances require a high degree of monitoring and continual testing.

35


leaving the program is of little value since a large majority of individuals had no drug tests conducted

in the years following Veterans Court exit. Unfortunately, even less testing occurred for the

comparison group in the two years following the disposition of their instant offense, rendering

comparisons of drug tests results between Veterans Court participants and the comparison group of

little value.

One limitation of this drug testing data is that it does not capture drug/alcohol tests administered by

the VA. Indeed, participants enrolled in certain VA treatment programs are subject to onsite drug

screening.33 Probation officers are privy to which of their clients the VA screens for substances while

active in formal VA treatment program. Given that 70% of participants who received treatment did

so through the VA (see below), it is possible that VA drug tests supplement the DOCCR drug

screening regimen and might greatly expand the number of participants tested for drugs.

Unfortunately, full VA drug test data is not available for all participants in the Veterans Court

evaluation sample because the signed participant agreements that permit the sharing of participant

data between the VA and the Veterans Court team have lapsed for the individuals in the evaluation

sample. As a result, the VA is no longer able to share participant data—which includes drug test

results—with the Hennepin County District Court Research Division.

However, some suggestive data exist that helps determine whether VA drug screening expands the

number of participants tested for substances. The Hennepin County Veterans Court program received

a federal Bureau of Justice Assistance grant from 2011 to 2014. One grant-reporting requirement was

submitting all drug test results from DOCCR and the VA on a quarterly basis. Examining the DOCCR

and VA drug test data for one of these quarterly grant reports reveals that a drug-testing gap remains

even after accounting for VA drug screens. Specifically, Figure 4 below documents the proportion of

participants active in the court between October and December 2012 who did and did not undergo

drug testing, and, if so, the agency (or agencies) administering the test. During these three months, the

33 In general, when participants test positive for illicit substances at the VA, the Veterans Justice Outreach Coordinator at Veterans Court notifies the Veterans Court team at the weekly pre-court staffing meetings. However, per conversations with the Veterans Justice Outreach Coordinator assigned to the Hennepin Country Veterans Court Program, the VA does not have a prescribed process or frequency for drug testing, except in formal group-based treatment programs (e.g. inpatient, outpatient, aftercare). If a veteran was only engaged in outpatient therapy, drug testing may never have occurred. Furthermore, the VA uses drug testing for treatment purposes versus compliance with court ordered conditions.

36


VA and DOCCR did not screen 42% of active participants for illicit substances, which similar to the

44% of Veterans Court participants in the evaluation sample who submitted zero drug/alcohol tests

during their program tenure. Of the participants tested by the VA or DOCCR during these three

months, 11 (26.8%) had fewer than three total drug tests, suggesting that the combined DOCCR and

VA testing regimen falls short of the truly random testing schedule stipulated by NADCP. Thus, while

adding the proportion of participants tested for substances by the VA increases the total number of

participants screened for drugs, the amount of participants not tested remains problematic.

Figure 4: Proportion of Veterans Court Participants Active in 4th Quarter 2012 Receiving Drug/Alcohol Test, by Testing Site (N=61)

Given the lack of a consistent—let alone random—drug testing protocol of Veterans Court

participants, determining whether the program was successful in its goal of facilitating participant

sobriety is difficult to judge. One measurable outcome was that for the individuals tested—which was

an equal proportion of graduates and non-completers—eventual graduates tested positive for

substances at significantly lower rates than eventual non-completers. However, DOCCR did not

screen over two-fifths of non-completers and graduates alike (44%), which is problematic because a

lack of testing does not indicate sobriety.

VA Only14% (N=10)

DOCCR Only27% (N=19)

VA & DOCCR17% (N=12)

No Tests42% (N=30)

37


Although the Veterans Court program fell short fully screening participants for drugs during its initial

years—which impaired the goal of facilitating participant sobriety—the program appears to be making

some progress with its drug-testing regimen. For example, 64 % of participants active in the program

as of October 2016 have had at least one drug test while 36% have not, which represents an increase

in participants screened for substances compared to the Veterans Court evaluation cohort. In the end,

because substance use disorders are a very common diagnosis among all Veterans Court participants,

it would strongly behoove the program to continue the recent trend of increasing the frequency of

drug tests, with the end goal of implementing a truly random testing regimen for all participants.

Goal 2: Increase compliance with treatment and other court ordered conditions

As mentioned above, the Veterans Court program has certain eligibility requirements, one of which is

that participants have a treatable behavioral or chemical health issue. This requirement is important in

light of the challenges veterans face: namely, the high rates of substance abuse and mental health

issues, which often occur in tandem in this population (see Baldwin 2016: 708-09). Given the

prevalence of these issues in the general veteran population, it us unsurprising that a majority of

participants enter the Hennepin County Veterans Court program with unmet mental and chemical

health needs. Table 23 below lists the different types of mental health diagnoses with which Veterans

Court participants present at program entry and exit. While 36 individuals do not initially have a

diagnosed mental health issue, many individuals present with more than one mental health diagnosis.

Because of the extensive array of services provided during the Veterans Court program—chief among

them treatment for chemical and behavioral health issues—many individuals have different or

additional diagnoses when they exit the program, noted on the right side of Table 22. This table shows

that the number of individuals without a diagnosis declined dramatically from entry to exit while the

number of individuals with multiple diagnoses rose. Most notably, there is a large jump in the number

of individuals diagnosed with a Substance Use and Addictive Disorder as well as the number of

participants with Post-Traumatic Stress Disorder. While it may seem as though the increase in

diagnoses at program exit is problematic, the identification of additional chemical or behavioral issues

allows individuals to get treatment that matches their needs, putting them on a more likely path toward

enhancing their mental health outcomes and increasing their overall quality of life after leaving the

program.

38


Table 22: Mental Health Diagnosis of Veterans Court Participants at Program Entry and Exit

At Program Entry At Program Exit

Diagnosis # Diagnosis #

Major Depression 69 Substance Use and Addictive Disorder 101 Post-Traumatic Stress Disorder 50 Post-Traumatic Stress Disorder 68 N/A or Unknown 36 Major Depression 67 Substance Use and Addictive Disorder 34 Generalized Anxiety Disorder 25 Generalized Anxiety Disorder 25 Traumatic Brain Injury 18 Traumatic Brain Injury 21 Bipolar Disorder 11 Bipolar Disorder 11 Other 11 Personality Disorder 6 Schizoaffective Disorder 7 Schizoaffective Disorder 6 N/A or Unknown 7 Schizophrenia 5 Borderline Personality Disorder 6 Other 5 Schizophrenia 6 Total 268 Total 327

Given the vast mental and chemical health needs with which Veterans Court participants present,

many participants undergo treatment by court order to address these issues. While not all participants

have a treatment program as part of their Veterans Court sentence, 106 out of 155 participants (68.4%)

went through a treatment program of some type. Of those who had treatment referrals, over 70%

attended treatment in a VA facility (typically at the VA Medical Center in Minneapolis or the VA

Hospital in St. Cloud, MN).

As Table 23 above demonstrates, the vast majority of participants (91.5%) referred to treatment

successfully completed the program; in fact, only nine individuals ordered to treatment failed to

complete successfully. In terms of treatment outcomes and overall program success, of the nine

participants who failed treatment, seven ended up not completing the Veterans Court program. Thus,

overall, it would appear that participants ordered to treatment programs are compliant with the

parameters of these treatment programs.

39


Table 23: Treatment Outcomes for Veterans Court Participants (N=106)


Did not complete treatment 7 2 9 (8.5%)

Completed treatment successfully 10 87 97 (91.5%)

One Veterans Court condition that is uniform across all participants is to remain law-abiding during

(and after) program participation. Table 24 shows the number of participants charged with a new

criminal offense during program participation. Only a fraction—less than 10%—of all participants

picked up a criminal charge during Veterans Court participation, suggesting that most participants

remained law-abiding in the strictest sense of the term. Graduates were less likely to garner new

criminal charges during program participation: 6% of eventual graduates received a new criminal

charge while active in Veterans Court (7 out of 119) and about 19% of non-completers had a new

criminal offense as an active Veterans Court participant (7 out of 36).

Table 24: New Criminal Charges during Veterans Court (N=155)


No new charges during 29 112 141 (91.0%)

At least one new charge during 7 7 14 (9.0%)

All seven eventual graduates ended up with convictions for the offenses they garnered during Veterans

court participation, while only four of the seven non-completers received a conviction for the new

criminal charge. While remaining law abiding is an important condition of Veterans Court, it is not

necessarily the case that a new criminal charge—or even a conviction—will derail a participant’s

chances of success on its own. Thus, the court plays an important role in helping participants handle

their legal issues in an ongoing and flexible manner, giving participants time and space to cultivate

meaningful behavioral changes.

40


Participants can also violate the spirit of being law abiding in ways that do not rise to the level of a

charge or conviction for a new criminal offense. Indeed, a more general way to determine whether

participants were law abiding during their participation is to see whether judges issued any warrants in

response to an infraction committed by a participant. Table 25 looks at the number of warrants issued

to active Veterans Court participants,34 breaking the sample into non-completers and graduates. As

this table shows, many participants—including a handful of graduates—had more than one warrant

issued during their participation. The data nevertheless suggest that a much higher proportion of

terminated participants had a warrant issued against them than graduates. Specifically, only 25% of

non-completers did not get a warrant for some type of violation during their participation in the

program (9 out of 36), whereas almost 75% program graduates did not commit an infraction that

resulted in a warrant (89 out of 119). It would thus appear that committing infractions that resulted in

the issuance of a warrant was detrimental to program success.

Table 25: Number of Warrants Issued during Veterans Court (N=155)


0 9 89 98 (63.2%)

1 9 18 27 (17.4%)

2 11 6 17 (11.0%)

3 3 4 7 (4.5%)

4 2 1 3 (1.9%)

5 1 1 2 (1.3%)

6 1 0 1 (0.6%)

Although the language of this goal—“an increase in compliance with treatment and court ordered

conditions”—suggests comparing compliance against a benchmark, no such metric exists. That said,

given the high rates of treatment success, the low rates of in-program offending, and the correlation

34 There are four types of warrants available to judges: Bench Warrants, Arrest, Complaint, Order of Detention, Bench Warrant-fail to appear at a hearing, and Violation/Probation Warrant. Each of these warrants is a response to a different type of infraction. Since individuals enter Veterans Court at different points in the criminal court process, all of these warrants are applicable to participants.

41


between compliance with program conditions and program success, the Hennepin County Veterans

Court program appears to be achieving this goal.

One limitation in assessing this goal is that the Hennepin County Veterans Court Program did not

track sanctions (or incentives) for the years when the participants in the evaluation sample were active.

Having data that are more robust on specific in-program sanctions would allow for a more direct

analysis of whether participants are complying with court rules and judge orders. It is highly

recommended that the Hennepin County Veterans Court team attempt to track sanctions for poor

behavior and incentives for good conduct systematically going forward.

Goal 3: Improve access to VA Benefits and Services

As mentioned above, it is not a requirement of the Hennepin County Veterans Court program for

participants to be eligible for VA services. Nevertheless, the vast majority of participants meet the

VA’s eligibility criteria to receive benefits, and about three-fourths (74.8%) of participants in Veterans

Court were at least minimally connected to VA benefits and services when they began the program.

However, as Table 26 below shows, almost half (49.7%) of Veterans Court participants saw an increase

in their VA benefit connectedness during Veterans Court, with the largest proportion being program

graduates. Furthermore, the percentage of participants not receiving benefits between program entry

and exit dropped to 7.7%.

Table 26: Change in Veterans Administration Benefit Connectedness during Veterans Court (N=155)


N/A, not receiving benefits 2 10 12 (7.7%)

Service level decreased 1 3 4 (2.6%)

Service level stayed the same 18 44 62 (40.0%)

Served level increased 15 62 77 (49.7%)

42


Given the jump in the number of individuals accessing benefits and the percentage of participants

increasing their service connectedness, the goal of improving access to VA Benefits and Services is an

unqualified success. A primary reason for this success is having representatives from the Veterans

Benefits Office and the Veterans Justice Outreach at the table working closely with the rest of the

Veterans Court team, which is instrumental to participants gaining access to benefits for which they

are entitled. A recommendation is to maintain the close links between the Veterans Court program

and multiple VA representatives.

Goal 4: Improve family relationships and social support connections

The principal way in which the Hennepin County Veterans Court team has sought to measure

participants’ progress on improving family relationships and social support connections is by looking

at participants’ number of prosocial activities—e.g. bowling leagues, recovery groups, volunteer

activities, etc.—and prosocial relationships—e.g. non-criminal friends, family, fellow service club

members, etc.

Table 27: Prosocial Activities at Veterans Court Entry and Exit (N=155) At Program Entry At Program Exit # (% of Total) # (% of Total)

Unknown 31 (20.0%) 2 (1.3%)

None 6 (3.9%) 13 (8.4%)

Minimal 53 (34.2%) 61 (39.4%)

Moderate 55 (35.5%) 61 (39.4%)

Many 10 (6.5%) 32 (20.6%)

After a referral to Veterans Court, the DOCCR screener makes a determination of the participant’s

levels of prosocial activities and relationships at Veterans Court entry based upon participant

interviews. At Veterans Court exit, probation officers determine their client’s levels of prosocial

activities and relationships based upon their knowledge of their client’s lives. Accurately capturing

improvement on these metrics is difficult as some of this data was not systematically captured in the

earliest phases of the program (and very difficult to reconstruct later), leaving no benchmark from

which to discern potential improvement for many participants.

43


Table 28: Prosocial Relationships at Veterans Court Entry and Exit (N=155)

At Program Entry At Program Exit # (% of Total) # (% of Total)

Unknown 30 (19.4%) 2 (1.3%)

None 15 (9.7%) 12 (7.7%)

Minimal 55 (35.5%) 47 (30.3%)

Moderate 47 (30.3%) 66 (42.6%)

Many 8 (5.2%) 28 (18.1%)

As Tables 27 and 28 indicate, about 20% of participants had no information for prosocial activities or

prosocial relationships at Veterans Court entry. That said, data was more complete at program exit,

and there is a clear increase in the instances of participants who have “moderate” or “many” prosocial

relationships and activities between program entry and exit.

Table 29: Change in Number of Prosocial Activities from Veterans Court Entry to Exit* (N=124)


Fewer prosocial activities at exit 9 16 25 (20.2%) Same number of prosocial activities at exit 9 41 50 (40.3%) More prosocial activities at exit 4 45 49 (39.5%)

*Participants whose level of prosocial activities and relationships was unknown at entry are not included in this table.

Tables 29 and 30 break down the trajectory of prosocial activities and relationships between entry and

exit, respectively (participants who did not have this data at program entry or exit were omitted).35

Table 30 demonstrates that while 20.2% of participants partook in fewer prosocial activities at

program exit than entry—including 16 graduates—the vast majority maintained their level of prosocial

activities or increased the number of prosocial activities in which they participated. Program graduates

35 Specifically, Table 29 omits 31 participants and Table 30 omits 30 participants.

44


comprised almost the entire group of participants who increased their array of prosocial activities (45

out of 49).

Table 30: Change in Number of Prosocial Relationships from Veterans Court Entry to Exit* (N=125)


Fewer prosocial relationships at exit 9 11 20 (16.0%) Same number of prosocial relationships at exit 7 42 49 (39.2%) More prosocial relationships at exit 7 49 56 (44.8%)

*Participants whose level of prosocial activities and relationships was unknown at entry are not included in this table.

Table 30, which examines the trajectory of prosocial relationships between program entry and exit,

largely mirrors the information in the previous table. Specifically, while some participants—including

11 graduates—had fewer prosocial relationships when they left the program, the bulk of participants

maintained or increased their circle of prosocial relationships. Graduates dominate the group that

increased their number of prosocial relationships.

All told, of those with prosocial activity and relationship data at entry and exit, graduates tended to

sustain or enhance their number of prosocial activities and prosocial relationships during program

participation. Terminated participants, on the other hand, were somewhat more likely to regress in

their number of prosocial activities and relationships. To the degree possible, the program should help

participants deepen their social connections and supports.

Table 31: Mentor Program Participation during Veterans Court (N=155)


Did not have Mentor 32 77 109 (70.3%) Had a Mentor 4 42 46 (29.7%)

One way in which Veterans Treatment Courts promote social supports is through mentor programs.

The first Veterans Court in the US—the Buffalo, NY program—created a mentor group to help

participants connect with veterans with similar military experience that serves as the template for

45


Veterans Treatment Court mentor programs around the country. According to the Buffalo program,

“The role of the veteran mentor is to act as a coach, guide, role model, advocate, and a support person

for the individual veteran participant with whom he/she is working” in the same manner an AA

sponsor might function.36 The mentor-mentee relationships have proved to be extremely popular with

and helpful to Veterans Court participants.

Unfortunately, the mentor program of the Hennepin County Veterans Court has never functioned in

a consistent manner, and the Court has had difficulty sustaining the mentor program.37 As a result,

there was not an equal opportunity for participants to receive a mentor during the courts initial years.

Of all participants in the evaluation, which represents some of the earliest participants in the Veterans

Court program, only 46 participants (29.7%) had a mentor at some point during their program

participation, as Table 31 shows. Yet, of the participants paired with a mentor, almost all completed

the program successfully (91.3%, or 42 out of 46), providing some suggestive evidence that

establishing a mentor-mentee relationship promotes success for Veterans Court participants.38

As of 2015, a new Mentor Coordinator has been on the Hennepin County Veterans Court team. While

the process of finding and training mentors has been going more slowly than anticipated, the mentor

program should ramp up in the latter half of 2016 and into 2017. Given that mentors can help

participants succeed in the program while providing important social supports, it is essential that the

Hennepin County Veterans Court program prioritize the rollout and continued maintenance of the

mentor program.

Despite the state of flux surrounding the mentor program, it does appear that the program is coming

close to meeting its goal of improving family relationships and social support connections, especially

given the wide increase in the number of prosocial activities and relationships from entry to exit. In

36 See: http://www.buffaloveteranscourt.org/content/buffalo-veteran-mentor-group 37 The Hennepin County Veterans Court mentor program was essentially moribund during 2014 and 2015. 38 Receiving a mentor is voluntary, yielding a concern that participants that are more social self-select into these relationships at higher rates. However, a cross tabulation between participants paired with a mentor and the number of prosocial relationships at program entry reveals that a plurality of participants paired with a mentor had “minimal” prosocial relationships when they began Veterans Court. As a result, it does not appear that highly social participants selected into the mentor-mentee relationship at higher rates than less social participants did.

46


addition, a newly revamped mentorship program should build upon the past successes the program

has had in this arena.

Goal 5: Improve life stability

In determining whether the program influences the life stability of participants, this study examines

housing, employment, and educational status at entry and exit.

Housing Status

A lack of stable housing is a longstanding problem among the US veteran population. Although

calculating the exact number of homeless veterans is difficult, estimates from 2010 suggest that 12,700

veterans of recent military conflicts were chronically homeless, with over 47,000 veterans homeless

on any single night (Baldwin 2016: 710). In Minnesota, about 8% of the state’s homeless population

is veterans, and six in ten homeless Minnesota veterans are chronically homeless (Wilder Research

2016). Given the housing instability facing the veteran population, the Hennepin County Veterans

Court program’s goal of improving housing status is laudable.

Table 32: Housing Status at Veterans Court Entry and Exit (N=155) At Program Entry At Program Exit Percent

Change # (% of Total) # (% of Total) Homeless 15 (9.7%) 14 (9.0%) -6.7%

Residential Facility 11 (7.1%) 14 (9.0%) 27.3% Relative/Friend 41 (26.5%) 27 (17.4%) -34.1% Independent 88 (56.8%) 100 (64.5%) 13.6%

Table 32 above shows the difference in housing status of Veterans Court participants at program

beginning and end. Importantly, 56.8% of Veterans Court participants were living independently at

the time they began the program. As a result, increasing housing stability was not an option for many

participants. That said, the percentage of participants who had independent housing at the time they

exited the program jumped to 64.5%. In addition, homelessness declined, but only marginally.

47


Figure 5: Housing Status Change from Veterans Court Start to Finish

Yet, only some Veterans Court participants enjoyed these gains in housing stability. Figure 5 above

displays the proportion of participants whose housing status decreased, stayed the same, or increased

from Veterans Court start to finish.39 While 66.5% all participants increased their housing stability

(103 out of 155), graduates accounted for the bulk of the increase. Furthermore, a much higher

proportion of terminated participants than graduates (39.9% to 10.9%) remained homeless or saw

their housing stability diminish from start to finish.

In the end, despite the overall progress made towards increasing housing stability, opportunities for

improvement exist. Specifically, the raw number of participants who were homeless at the beginning

of the program (15) was hardly different from the number of homeless at program exit (14). Although

homelessness does not affect a large number of participants, it does seem to be difficult to rectify—

39 Housing stability decreases are defined as participants who homeless at entry and exit or participants whose housing stability decreased (e.g. from living independently to living with a relative/friend). Housing stability is labeled as staying the same at entry and exit if nothing changed from beginning to end, or if there was a housing status change that did not result in living independently or becoming homeless. Housing stability is categorized as increased for participants who lived independently from program beginning to end, participants who gained independent housing, and participants who stopped being homeless.

47.2% (N=17)

72.3% (N=86)66.5% (N=103)

13.9% (N=5)

16.8% (N=20)16.1% (N=25)

38.9% (N=14)

10.9% (N=13)17.4% (N=27)

0.0%

20.0%

40.0%

60.0%

80.0%

100.0%

Non-Completers Graduates All

Increased Housing Status Housing Stayed the Same Still Homeless or Decreased Housing Stability

48


even for program graduates—and more attention and resources should be devoted to individuals who

enter the program without housing or those living in unstable living situations. In addition, considering

that graduates enjoy better housing outcomes than non-completers, the Veterans Court team should

help ensure that all participants have the opportunity to enhance their housing status. Such efforts

would help end the scourge of homelessness among the veteran population.

Employment Status

Similar to unstable housing, a lack of employment is another difficulty facing veterans. For example,

“Gulf War-era veterans were unemployed at higher rates than nonveterans in 2013 and 2014”

(Baldwin 2016: 710). Fortunately, the program was largely successful in enhancing employment, as

Table 33 shows. At the start of the Veterans Court program, over a quarter (27.1%) of participants

had full-time employment, whereas around a third (31.0%) had full-time employment at exit, a 14.3%

increase. Part-time employment ticked up 55.6% from beginning to end. Furthermore, 10 additional

participants (a 33.3% jump) were qualified as disabled—giving them access to disability benefits—at

Veterans Court end compared to Veterans Court start. Most impressively, unemployment dropped

dramatically, from 45 participants without work at program entry to 24 at program exit, which is a

46.7% drop.

Table 33: Employment Status at Veterans Court Entry and Exit (N=155) At Program Entry At Program Exit Percent

Change # (% of Total) # (% of Total) Unemployed 45 (9.7%) 24 (15.5%) -46.7%

Disabled 30 (7.1%) 40 (25.8%) 33.3% Retired 6 (3.9%) 6 (3.9%) 0.0% Student 9 (5.8%) 8 (5.2%) -11.1% Part-time 18 (11.6%) 28 (18.1%) 55.6% Full-time 42 (27.1%) 48 (31.0%) 14.3% Unknown 5 (3.2%) 1 (0.6%) -80.0%

49


Once again, program graduates saw the majority of these gains. Figure 6 shows that over half (52.1%)

of graduates increased their employment status while only 8.4% of graduates saw their employment

status decrease.40 The opposite was true for non-completers, whereby decreasing employment status

(38.9%) was more common than increased employment stability (13.9%). That said, it was most

common for non-completers to have the same employment status from start to finish (47.2%).

Figure 6: Employment Status Change from Veterans Court Start to Finish

Although there is progress overall, opportunities for improvement exist. First, despite the drop in

unemployment among graduates, the fact that 8.4% of successful program completers were

unemployed throughout their participation or became unemployed during the program indicates that

there is still work to do even among those that succeed in Veterans Court. In addition, the Veterans

Court team must work to ensure that all participants see gains in their employment statuses, and not

just the graduates.

40 Participants who maintained full-time employment throughout the program or increased their employment status to full- or part-time compose the “Increased Employment Status” group. Participants who maintained part-time employment or another non-unemployed status throughout participation, became a student, or got on disability compose the “Stayed the Same” group. Employment decreases are defined as being unemployed throughout participation or becoming unemployed during participation.

13.9% (N=5)

52.1% (N=62)43.2% (N=67)

47.2% (N=17)

39.5% (N=47

41.3% (N=64)

38.9% (N=14)

8.4% (N=10)15.5% (N=24)

0.0%

20.0%

40.0%

60.0%

80.0%

100.0%


Increased Employment Status Employment Stayed the Same Unemployed or Stayed Unemployed

50


It bears mentioning that the gains in employment are even more impressive given that contact with

the criminal justice system often impedes employment prospects among criminal offenders (O'Brien

and Darrow 2007). Fortunately, new initiatives should help expand the court’s success in expanding

employment for justice-involved veterans, such as Minnesota’s recent “ban the box” law that prohibits

employers from asking about criminal history on job applications. With the Minnesota Department

of Employment and Economic Development (DEED) as a member of the Veterans Court steering

committee, it is hopeful that Veterans Court participants will continue making progress toward

enhancing their employment prospects as part of their participation in the program.

Educational Status

Unlike housing and employment, educational status is much less fluid, and change only happens in

one direction. Specifically, high school and college graduates do not lose degrees or diplomas

conferred upon them. In addition, earning a degree tends to take longer than the 12-18 months

participants spend in the program. Furthermore, many higher education programs have restrictive

times when new or returning students can begin coursework (e.g. fall semester). Moreover, individuals

employed on a full- or part-time basis might not have the flexibility in their work schedules to

accommodate taking classes. With these caveats in mind, it is unsurprising that the vast majority

(81.9%) of Veterans Court participants saw no change in their levels of education from program start

to finish as Figure 7 below shows.

Table 34: Education Status at Veterans Court Entry and Exit (N=155) At Program Entry At Program Exit Percent

Change # (% of Total) # (% of Total) Less than High School 6 (3.9%) 3 (1.9%) -50.0%

High School Diploma/GED 52 (33.5%) 45 (29.0%) -13.5% Some College 59 (38.1%) 53 (34.2%) -10.2% Two-Year Degree 20 (12.9%) 28 (18.1%) 40.0% Four-Year Degree and Up 18 (11.6%) 26 (16.8%) 44.4%

Yet there were plenty of educational status gains. Specifically, Table 34 above demonstrates that many

participants increased their educational status during Veterans Court participation in one way or

51


another. There were eight individuals who received four-year degrees during their Veterans Court

tenure, another eight who earned a two-year degree, and three individuals who got a GED. Not

surprisingly, 16 of the 19 individuals who increased their level of education were program graduates.

Figure 7 shows that a further nine participants (seven graduates and two non-completers) began taking

college courses during their time in Veterans Court. Given the impediments to beginning and

maintaining coursework while in the Veterans Court program, the fact that many individuals were able

to increase their educational status suggests that the program is indeed helping participants increase

their level of education, which will likely have positive downstream effects for employment prospects

and housing stability.

Figure 7: Education Status Change from Veterans Court Start to Finish

All told, upon examining participants’ changes in employment, housing, and education from program

start to finish, it appears that many participants are improving their life stability in accordance with

the goals of program. Yet, these goals are not widely distributed across all program participants and

the gains tend to be concentrated among graduates. While it is not surprising that graduates enjoy the

lion’s share of the program’s successes, Veterans Court needs to work for all participants, and not just

the graduates.

86.1% (N=31) 80.7% (N=96) 81.9% (N=127)

5.6% (N=2)5.9% (N=7) 5.8% (N=9)

8.3% (N=3) 13.4% (N=16) 12.3% (N=19)

0.0%

20.0%

40.0%

60.0%

80.0%

100.0%


No Education Change Started Ed Program During VC Earned Degree during VC

52


Goal 6: Reduce Criminal Recidivism

The last goal of the Veterans Court program that this study evaluates is whether the program reduces

participants’ contact with the criminal justice after the program. Evaluating this goal focuses on the

“treatment effect” of the Veterans Court program. As such, it is insufficient to compare graduates to

non-completers as both groups received the treatment. Instead, effectively addressing this goal

requires comparing the cohort of Veterans Court participants to the group of individuals who are

similar to our Veterans Court participants on many important levels, but did not receive the treatment

as described above. Given that the two groups were matched upon all available variables (see Tables

15-19), difference of means tests are sufficient to determine differences in recidivism.

Before the recidivism analysis, there are several important definitions and caveats. First, it is possible

that certain attributes not included in the array of matched variables could affect subsequent criminal

activity in unobserved ways. Specifically, there is no military service information or life quality

measures for the comparison group, and it is possible these groups differ on these attributes in a

systematic manner. Although these unobserved differences could account for divergent outcomes

between the Veterans Court cohort and the comparison group, the legal and extralegal variables used

to match these two groups provide a solid basis for assuming these two groups recidivism are

“equivalent and unbiased,” as prescribed by NADCP 2015.

This study analyzes new criminal activity during a two-year recidivism window. The two-year window

for Veterans Court begins on the day the individual leaves the program (either successfully or

unsuccessfully). While some Veterans Court evaluations include the time a participant spends in the

program as part of the recidivism window (e.g. Hartley and Baldwin 2016), this study does not for two

reasons. First, as noted above, the Hennepin County Veterans Court program has policies and

procedures to handle new criminal offenses committed during participation. Second, there is a major

incongruence between the high levels of supervision participants receive while in the program and the

absence of this supervision afterwards, and it is most valuable to understand how participants function

once they leave the highly structured Veterans Court environment. As a result, it is ideal to use the

end of each participant’s tenure in Veterans Court as the beginning of the 730-day recidivism window.

53


For the comparison group, two-year follow-up period begins on sentence date of their instant offense.

Because new criminal activity or probation violations on existing offenses can lead to jail time where

individuals are physically incapable of recidivating, the two-year window takes into account the

number of days an individual spends in jail/prison during this time, such that all participants have a

full 730 days of non-incarcerated “street time” in which to recidivate.41 NADCP’s (2015) best practice

standards for treatment court evaluations suggest a three-year recidivism window. However, due to

the relative youth of the Hennepin County Veterans Court combined with the 12-18 months it takes

to finish the program, there are not a sufficient number of former Veterans Court participants with

that length of a recidivism window after considering “street time.” Given the choice between analyzing

more participants or a longer recidivism window, this study prioritizes the former.

Finally, the term recidivism can take on different meanings and measures, some of which are not

appropriate for this study. For example, some problem solving court evaluations define recidivism as

new arrests (e.g. Hartley and Baldwin 2016). Comparing arrests can be misleading because of

inconsistent policies across different cities.42 In addition, studies show higher arrest statistics for

minority groups even when self-reported offending is similar to, or less than, that of white offenders

(see ACLU 2014). As a result, using new arrests as a measure for recidivism is not an ideal choice.

Another possible definition of recidivism is a new criminal charge. One benefit is that formal criminal

charges weed out arrests with no subsequent legal action. Yet, similar to arrests, criminal charges tend

not to be distributed in an equitable manner and end up with systematically different dispositional

outcomes. Johnson (2015), for example, found that African Americans are much more likely than

Whites are to have their charges dismissed in Hennepin County, suggesting that prosecutors charge

Black defendants more often with cases that ultimately cannot stand up in court. As a result, using

charges as the measure of recidivism would likely affect different groups disproportionately, regardless

of the legal outcome of the case. Most egregiously, equating charges with criminal activity potentially

41 Some individuals in both samples ended up with long prison terms that precluded them from having a full two years of street time. In these instances, the cutoff date for the recidivism window was June 1, 2016. Despite the fact that some individuals had less than two years of street time because of long prison sentences, these prison terms stemmed from convictions received during the recidivism window. Thus, the recidivism analysis adequately captures subsequent criminal activity even among offenders with less than 730 days of street time. 42 There are dozens of police agencies in Hennepin County, each with different policies to adhere to and ordinances to enforce.

54


violates the spirit of “innocent until proven guilty,” which serves as the foundation of the United

States’ criminal justice system.

Given these limitations, this evaluation uses new convictions to test for recidivism outcomes in the

hopes that it will minimize potential sources of implicit bias and accurately reflect actual criminal

activity. More directly, this study defines recidivism as a new conviction during the two-year follow

up window.43

Figure 8: Percent of Veterans Court and Comparison Group who Recidivated

In the most general of terms, fewer members of the Veterans Court cohort recidivated than the

comparison group, as Figure 2 above demonstrates. Specifically, the recidivism rate (the proportion

of each group that recidivated, see Hartley and Baldwin 2016: 13) for the Veterans Court sample is

22.6%, while the recidivism rate for the comparison group is 26.5%. However, this difference does

not rise to statistical significance, suggesting no demonstrable difference in criminal activity between

the two groups.

43 This analysis omits petty misdemeanor and payable misdemeanor traffic offenses, such as Driving after Revocation.

77.4%(N=120)

22.6%(N=35)

73.5%(N=114)

26.5%(N=41)

0

10

20

30

40

50

60

70

80

90

100

No new conviction At least one new conviction

Veterans Court Comparison Group

55


Table 35: Recidivism Analysis between Veterans Court Cohort and Comparison Group

Subsequent Misdemeanor Convictions Mean Minimum Maximum Std. Dev. Comparison Group 0.4 0 4 0.8


Subsequent Gross Misdemeanor Convictions Mean Minimum Maximum Std. Dev. Comparison Group 0.1 0 4 0.5 Veterans Court 0.1 0 1 .3

Subsequent Felony Convictions Mean Minimum Maximum Std. Dev. Comparison Group 0.1 0 3 0.4 Veterans Court 0.1 0 3 0.4

Total Number of Subsequent Convictions Mean Minimum Maximum Std. Dev. Comparison Group 0.6 0 5 1.2 Veterans Court 0.5 0 10 1.4

Any Subsequent Conviction (Yes or No) Mean Minimum Maximum Std. Dev. Comparison Group 0.3 0 1 0.4 Veterans Court 0.2 0 1 0.4

Recidivism Points Mean Minimum Maximum Std. Dev. Comparison Group 0.8 0 13 2.0 Veterans Court 1.2 0 21 3.1

Table 35 above shows that differences in new convictions between the Veterans Court cohort and the

comparison group are not statistically significant, regardless of the measure of recidivism. In fact, the

Veterans Court group averaged a higher number of recidivism points than the comparison group in

the two-year follow-up window (mainly because the Veterans Court group received more felony

convictions, see Figure 10 below). Nevertheless, as mentioned, none of the differences in the ways to

56


measure recidivism rose to statistical significance, indicating that Hennepin County Veterans Court is

falling short of its goal of reducing criminal recidivism.

Figure 9: Time to Recidivate for Veterans Court Participants and the Comparison Group

Despite the fact that Veterans Court participants do not recidivate at lower rates than the comparison

group, it is worth noting different patterns of recidivism between these groups. Figure 9 below shows

the time (in days) to recidivate for each group. While the overall recidivism rate is higher for the

comparison group, Veterans Court participants have a higher recidivism rate in the first 120 days of

street time. Specifically, the Veterans Court group recidivates more often in the first few months of

street time. In contrast, the comparison group’s recidivism rate jumps dramatically between 120 and

180 days, when it leapfrogs the Veterans Court cohort’s recidivism rate.44 Considering that many

Veterans Court participants recidivate quickly, one policy recommendation is providing extra

44 One possibility for the recidivism jump in the comparison group between 120 and 180 days of street time are changes in probation supervision levels. Unfortunately, court data cannot account for changes in probation supervision levels.

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

30.0%

30 60 90 120 150 180 210 240 270 300 330 360 390 420 450 480 510 540 570 600 630 660 690 720

Reci

divi

sm R

ate

Days to Recidivate


57


probation supervision to individuals who have just left Veterans Court, in the hopes that additional

probation supervision will help reduce the risk of reoffending during the critical first several months.

Figure 10: Total Number of New Convictions, and New Convictions by Type

Another consideration is the type and number of new convictions for each sample. Figure 10 below

documents the total number of new convictions garnered in the two year follow up period, which

displays some differences between the Veterans Court cohort and the comparison group. The data

show that during the two-years of street time, the individuals in the comparison received a handful of

more convictions than the Veterans Court group (88 to 80), a difference that was not statistically

significant. Drilling down, Veterans Court participants garnered several more felony convictions than

the comparison group, but fewer gross misdemeanors and common misdemeanors (differences that

were not statistically significant).

Although the full Veterans Court cohort did not commit fewer new crimes than the matched

comparison group, looking exclusively at graduates of the program might produce differenc results.

Hartley and Baldwin (2016) found that Veterans Court graduates recidivated less than the comparison

identified by the authors. However, Hartley and Baldwin compared graduates to their full comparison

sample, which is not appropriate as program graduates could differ in systematic ways from the full

80

1712

51

88

11

20

57

0

30

60

90

Total number of newconvictions

Number of new felonyconvictions

Number of new grossmisdemeanor convictions

Number of newmisdemeanor convictions


58


comparison group. To ensure that Veterans Court graduates receive a proper comparison group, this

study once again used propensity score matching to pair Veterans Court graduates with comparison

group matched on race, age, gender, criminal history, and instant offense type.45

Table 36: Recidivism Analysis between Veterans Court Graduates and Comparison Group

Subsequent Misdemeanor Convictions Mean Minimum Maximum Std. Dev. Comparison Group 0.3* 0 4 0.7

Veterans Court 0.1* 0 2 0.3

Subsequent Gross Misdemeanor Convictions Mean Minimum Maximum Std. Dev. Comparison Group 0.1 0 3 0.4 Vet Court Grads 0.1 0 1 0.2

Subsequent Felony Convictions Mean Minimum Maximum Std. Dev.

Comparison Group 0.1 0 4 0.4 Vet Court Grads 0.1 0 2 0.3

Total Number of Subsequent Convictions Mean Minimum Maximum Std. Dev.

Comparison Group 0.5* 0 8 1.1 Vet Court Grads 0.2* 0 2 0.5

Any Subsequent Conviction (Yes or No) Mean Minimum Maximum Std. Dev.

Comparison Group 0.3* 0 1 0.4 Vet Court Grads 0.1* 0 1 0.3

Recidivism Points Mean Minimum Maximum Std. Dev. Comparison Group 0.7 0 17 1.9 Vet Court Grads 0.4 0 6 1.2

* Statistically significant at the .01 level

45 Once again, this study used a one-to-many match was used to Veterans Court graduates. There were 79 unique comparison group individuals yielding 119 observations. There were no statistically significant differences between Veterans Court graduates and the comparison group on the matched variables.

59


As Table 36 shows, there are some statistically significant recidivism differences between Veterans

Court graduates and its matched comparison group. Principally, the average number of subsequent

misdemeanor convictions, the average number of total convictions, and the proportion of individuals

with a subsequent conviction were all lower for the Veterans Court graduate groups, differences that

were all statistically significant. However, there were no statistically significant differences between

these groups when looking at subsequent gross misdemeanor convictions, subsequent felony

convictions, or the total number of recidivism points. Thus, the differences in recidivism between

these groups largely revolved around the fact that the comparison group garnered more low-level

convictions than did Veterans Court graduates. Nevertheless, the data clearly demonstrate that

Veterans Court graduates reoffend less than a cohort of justice-involved veterans with similar profiles.

Figure 11: Recidivism among Veterans Court Graduates and Non-Completers

Successfully graduating from the program appears to produce beneficial post-program outcomes,

which contrasting graduates against non-completers further highlights. Furthermore, Figure 11 above

shows that the recidivism rate is much higher among individuals terminated from the program

compared to graduates (58.3% to 11.8%). Program graduates reoffended less at statistically significant

rates across all measures of recidivism compared to non-completers (data not shown). Given that

88.2%(N=105)

11.8%(N=14)

41.7%(N=15)

58.3%(N=21)

0

10

20

30

40

50

60

70

80

90

100

No Conviction At least one conviction

Graduates Non-Completers

60


there is no difference in recidivism between the Veterans Court cohort and the comparison group for

all Veterans Court participants and statistically significant differences in reoffending between program

graduates and that comparison group, the Veterans Court program must double down on helping all

participants graduate, which will likely have positive spillover effects on achieving program goals.

In sum, while the Veterans Court program is largely unsuccessful in the narrowly tailored goal of

reducing criminal recidivism when examining all participants, there were some important lessons.

First, when former Veterans Court participants recidivate, most tend to do so quickly, which has

implications for supervision levels after participants leave the program. Second, graduates of the

program do indeed recidivate less, albeit only at the lowest charge level. Finally, graduates recidivate

significantly less than individuals who terminate from the Veterans Court program. Since graduation

curtails recidivism, it bears taking a deeper look into what makes success in Veterans Court more

likely.

The Factors that Make Veterans Court Success More Likely

Given the post-program conviction discrepancies between Veterans Court graduates and non-

completers as well as the recidivism differences between program graduates and the statistically-similar

comparison group, discovering the factors that make program success more or less likely could help

graduate more individuals from Veterans Court in the future (i.e. through identifying areas where

additional services could be offered). In turn, this could make the Veterans Court program more

effective at reducing criminal recidivism since graduates have better outcomes on that metric.

In determining what makes success in Veterans Court more likely, this study relies on logistic

regression analysis looking only at program participants. Logistic regression allows for the inclusion

of numerous variables to determine which ones affect program success while holding the remaining

variables constant; that is, this method isolates the individual effect of each variable on program

success independently of one another.

Choosing which variables to include in this regression analysis is somewhat difficult as there are

immutable and dynamic factors at play during each participant’s Veterans Court tenure. Whereas some

factors will not change over the course of program participation—such as race/ethnicity, criminal

61


history, and military service—other factors will change, in part because the program targets certain

areas—such as housing or employment status. In addition, participant behavior in the program is also

important to consider—for example, whether participants test positive for illicit substances or choose

to have a mentor. The regression model below includes several categories of variables to account for

the factors that likely impact program success while accounting for these dynamic and fixed participant

attributes. A first array of variables is demographic, and includes age and race/ethnicity.46 This model

excludes gender as there is minimal variation—only 3% of the sample is female.

The second set of variables relates to the criminal profile of each participant, and includes criminal

history points and the level of the instant offense (specifically whether the charge was a gross

misdemeanor).

The third set of indicators are military-specific variables, specifically whether the participant exited the

military at a rank of E3 or below and whether the participant’s discharge status was anything less than

“honorable,” with the presumption that a low rank or discharge status might impede program success.

The fourth set of variables consists of three quality of life measures: whether the participant began the

program with high education,47 whether the participant had a low housing trajectory,48 and whether

the participant had a low employment trajectory.49

The final array of indicators centers on in-program events and participant behaviors. This model

considers whether the participant had a mentor, whether the participant spent any time in jail on their

instant offense during program participation,50 whether a judge issued any warrants against the

participant during program participation,51 and whether the participant had any positive drug tests.

46 We also include the squared value of age in the model. Adding the square of the age allows the model to test the effect of age—which may have a nonlinear relationship with the dependent variable—more accurately. For example, it is possible that age has a negative relationship with Veterans Court success up to a certain point, and a positive relationship thereafter. Given the tendency of criminogenic tendencies to change as individuals grow older (see Loeber and Stallings 2011), modeling age as a potentially nonlinear variable by including its squared value is germane. 47 Defined as whether the participant had already completed a two- or four-year degree. 48 Defined as whether the participant was homeless from start to finish or became homeless during participation. 49 Defined as whether the participant was unemployed from start to finish or became unemployed during participation. 50 Which could be a sanction from the judge for some infraction of program rules. 51 Which might indicate a formal response to a serious breach of program rules.

62


The results of the model are in Table 37. Several variables are statistically significant in the model

(denoted with an asterisk), suggesting that some indicators have an impact on program success.52

Table 37: Determinants of Veterans Court Success

Independent Variables

Coefficient

Std. Error Significance

Level Odds Ratio

Nonwhite -0.630

0.56

0.258

0.533 Age 0.258

0.18

0.151

1.195 Age Squared -0.003

0.00

0.208

0.997 Criminal History Points* -0.096

0.04

0.015

0.909

Gross Misdemeanor Instant Offense* 2.183

0.66

0.001

8.876 Exited Military at Low Rank -0.311

0.59

0.598

0.732

Less than Full Honorable Discharge -0.433

0.80

0.590 0.649 Low Housing Status* -0.535

0.63

0.005

0.586

Low Employment -2.190

0.77

0.396

0.112 High Education at Start 0.277

0.83

0.738

1.318

Mentor During VC* 2.073

0.84

0.014

7.948 Some Jail Time Served during VC 0.456

0.73

0.534

1.578

Warrant Issued During VC* -1.057

0.44

0.017

0.348 Positive Drug/Alcohol Test -0.329

0.58

0.571

0.720

* Statistically significant variable; Pseudo R2 = .4133; N=155

A first finding of the model is that the level of charge matters for program success. Participants who

enter the program facing gross misdemeanor charges are far more likely to graduate from the program

as compared to participants facing felony or common misdemeanor charges.53 The odds ratio statistic

in the final column of the table suggests that participants facing gross misdemeanor charges are almost

9 times more likely to graduate than are participants facing felonies or common misdemeanors. In

another regression analysis (not shown) the results indicate that having a misdemeanor as an instant

offense charge is detrimental to program success compared to a felonies or gross misdemeanors,

whereby participants with misdemeanor instant offenses are 46% more likely to fail than individuals

facing felony or gross misdemeanor level instant offenses. Interestingly, there is no statistically

significant effect of entering the program with felony level instant offenses, despite this being the most

52 The Pseudo R-square value of .4133 suggests that the variables included in the model explain about 41% of the variation in program success. 53 As Table 15 above shows, 55 of the 70 gross misdemeanor instant offenses (78.6%) were for Gross Misdemeanor DWIs.

63


serious level of charge. The statistically significant effect for some instant offense types has important

implications, whereby gross misdemeanant participants likely require less supervision and

programming than other participants do. The greater likelihood of failure for participants facing

misdemeanor instant offenses might revolve around the fact that the rigors of the Veterans Court

program do not compare well with typical sentences for low-level offenses. We discuss the

implications of this finding in more detail in the recommendation section.

A second variable that rose to statistical significance is criminal history, whereby a greater number of

criminal history points make program failure more likely. More directly, each additional criminal

history point increases the odds of program failure by about 10%. Thus, compared to a participant

with no criminal history, an individual with a prior “person felony” conviction is 40% less likely to

graduate from the program (see above for an explanation for calculation and definition of criminal

history points). Given the links between an individual’s prior convictions and program outcome, the

Veterans Court team should provide varying levels of supervision and programming commensurate

with a new participant’s criminal history.

Individuals who are homeless throughout their entire tenure in Veterans Court or who become

homeless during the program are 71% more likely to terminate. This was the only “life quality” metric

that rose to statistical significance in the regression analysis; that is, employment and education status

had no statistically significant relationship with program outcome. The Veterans Court team should

devote extra resources to homeless veterans entering the program to help them find stable housing

while ensuring that active participants maintain their housing.

One variable capturing in-program participant behavior rose to statistical significance: having a

warrant issued while an active participant. Participants who have a warrant issued against them during

their participation in Veterans Court are almost three times less likely to graduate. A warrant issued

against a Veterans Court participant likely demonstrates that the participant is out of compliance with

program rules, suggesting suboptimal program progression. Importantly, warrants do not lead to

program failure automatically (According to Table 25, 30 out of 119 eventual graduates received a

warrant while an active Veterans Court participant). Rather, warrants are red flags. When a participant

behaves in a way that merits the issuance of a warrant, it is likely a warning sign that they are at risk of

64


failure, and the team should take steps to work with the participant on correcting behavior after they

receive a warrant.

Finally, that last statistically significant independent variable, and one of the most potent predictors of

success was whether the participant had a mentor during Veterans Court participation. Controlling

for all other factors, having a mentor makes a participant almost eight times more likely to graduate

from the program. By virtue of the shared experience of being in the US Armed Forces, mentors can

assist Veterans Court participants in a manner that is difficult for other members of the Veterans

Court team. Thus, a clear recommendation from this study is that the Hennepin County Veterans

Court program must quickly ramp up its mentor program as soon as possible, which would also allow

the program achieve its goal of connecting participants to social supports.

This regression model suggests several factors enhance or inhibit participant success. It appears that

some participants are more primed for success right away. Specifically, participants with less extensive

criminal histories facing gross misdemeanor charges who are not homeless tend to be more likely to

graduate. Yet, events that occur during the program matter as well: receiving a warrant imperils a

participant’s chances of success while having a mentor greatly boosts the prospects of graduation.

RECOMMENDATIONS

Many recommendations follow from the evidence uncovered in this evaluation. First, given that

graduates have better post-program outcomes than non-completers, a main goal of the program going

forward should be to graduate more participants. Of course, the designation of “graduate” does not

improve participant outcomes on its own, and watering down the graduation criteria in the hopes of

graduating more participants would be counterproductive. Rather, the behavioral changes the program

helps foster serve as the foundation for the enhanced outcomes graduates enjoy. This evaluation of

the Hennepin County Veterans Court program identified several explicit areas in which the program

can improve participants’ likelihood of graduation.

First, participants who had a mentor were far more likely to complete the program successfully than

individuals without a mentor. Despite the success of the mentor program, which assisted the Veterans

Court program in its goal of improving family relationships and social support connections, the

65


mentor program effectively ceased functioning for several years. Thus, a first major recommendation

would be to expand the newly revamped mentor program as soon as possible.

Furthermore, the program should be judicious with which participants receive a mentor. If mentors

are in short supply, the program should pair participants most at risk of failure with mentors. The data

analysis above identified several participants attributes associated with program failure. For example,

individuals referred to the program with more extensive criminal histories are more likely to fail, and

can be given higher levels of supervision immediately. In addition, since participants with gross

misdemeanor instant offenses were far more likely to succeed than participants facing felonies or

misdemeanors, gross misdemeanants likely need much less supervision and fewer services, especially

those with no criminal history and stable housing. Tailoring levels of supervision and services to a

participant’s needs will help stretch the pool of resources at the Veterans Court team’s disposal.

While gross misdemeanor instant offenses were associated with higher rates of graduation, participants

facing misdemeanor level charges were prone to failure. One potential reason for the high failure rates

of misdemeanants is that the rigors of Veterans Court might appear more expensive in terms of time

and effort than the typical criminal sentence for a misdemeanor, causing these participants to drop

out of the program in favor of the traditional criminal court process. Yet, these individuals might be

in desperate need of the program’s services despite the low-level charge. In order to discern which

misdemeanants are most appropriate for the program, the Veterans Court team should include a

validated risk/needs assessment during participant screening. Indeed, in an evaluation of a Veterans

Court that accepted only misdemeanants, Hartley and Baldwin (2016) found that higher criminogenic

risk levels affected recidivism.54 Ideally, participants would receive an assessment at program entry to

ensure the program understands their risk level and serves their needs.

Housing status was another factor that influenced graduation rates. Specifically, individuals who were

homeless from beginning to end or became homeless during program participation had a higher

likelihood of terminating from the program. In the years since the Hennepin County Veterans Court

program began, the US government has instituted goals and programs to end homelessness among

the veteran population, which represents a helpful starting point. Yet, because not all Hennepin

54 This jurisdiction used the Wisconsin Risk/Needs Scale to measure risk and need.

66


County Veterans Court participants are eligible for VA services, the Veterans Court team should add

a community-based housing advocate to the Steering Committee so that participants can get housing

services immediately if they are homeless or have a precarious housing situation.

One aspect that has not received much attention thus far is the very low graduate rate of Native

American participants. Although they do not represent a large proportion of participants—a “small

sample size” to use common parlance—five of the six Native American participants evaluated in this

study did not complete the program. In contrast, the vast majority of participants in every other

racial/ethnic group in the study successfully graduated. Community-based providers that offer

culturally specific services to Native American participants who strongly identify with Native culture

might be more effective than traditional VA services. Alternatively, the VA has an Office of Tribal

Government Relations that “consults with Native American and Alaska Native tribal governments to

develop partnerships that enhance access to services and benefits by Veterans and their families.”55

A recommendation of this study is for the VA liaisons on the Veterans Court team to study whether

the Office of Tribal Government Relations can provide culturally specific services or help link

participants to culturally competent service providers.

The recidivism analysis showed that many Veterans Court participants recidivated quickly upon

leaving the program. One way to reduce reoffending would be to lessen supervision levels after

participants leave the program only gradually. During program participation, participants are subject

to high levels or supervision and expectations, much of which ceases after leaving the program.

Instituting a graduated scale that maintains some of the program structure in the first couple of months

after a participant leaves the program and then tapers off in the subsequent months might help

participants remain law abiding in the short and long term. Fortunately, the program recently began

holding post-program review hearings for participants who have left the program, although it is too

early to discern whether these post-program hearings have affected recidivism rates.

Finally, the program can easily adopt changes to move closer to NADCP’s best practice standards for

treatment courts. As mentioned in the section evaluating the program’s sobriety goal, the program is

not screening participants for illicit substances in a systematic manner. Specifically, upwards of 44%

55 See: http://www.va.gov/tribalgovernment/

67


of the participants in this evaluation received no DOCCR-administered drug tests while active in the

program. Although there is suggestive evidence that DOCCR is screening a greater number of

participants now, the Veterans Court team should adopt a truly random drug testing policy for all

participants. Furthermore, the program has never had any procedures in place to capture incentives

and sanctions. Tracking rewards and penalties stemming from participant behavior could help the

program enhance its behavioral modification regimen.

These recommendations are the product of a rigorous program evaluation, and could help the next

wave of veterans succeed in the program while helping the program meet its goals.

CONCLUSION

Hennepin County initiated its Veterans Court program in 2010. Since then, the program has served

hundreds of clients. The Hennepin County Veterans Court enjoys an admirable graduation rate

relative to other treatment courts, yet judging its success requires examining whether the program is

meeting the goals it set for itself. The scorecard in Table 38 below reflects the findings of the thorough

evaluation of program’s goals and highlights some of the successes and shortcomings.

Starting with the goal of enhancing participant sobriety, success was difficult to discern because of the

relative lack of drug testing that took place. The program should work to increase the number of drug

tests participants are required to submit during and after the program by adopting a truly random

drug-testing regimen. While the rates of positive drug tests in the program are not high, over two-

fifths of participants never had to take a drug test, making it difficult to determine the degree to which

the program actually facilitates participant sobriety.

The program fared better on the goal of increasing compliance with treatment and other court ordered

conditions. Most participants were engaged in at least one formal behavioral health treatment

program, and the vast majority of these individuals successfully completed the treatment referral.

Furthermore, due to the additional access to mental and chemical health services, participants tend to

have more robust and accurate chemical and behavioral health diagnoses at program exit, helping

them to receive the correct treatment regimen going forward. Finally, few participants picked up new

criminal charges while active in the program, indicating compliance with the program’s behavioral

expectations.

68


The most successful program goal was helping participants improve their access to VA benefits and

services, whereby over half of participants saw their access to VA benefits increase during program

participation (the majority of whom the VA already served). Furthermore, the percentage of

participants not receiving benefits between program entry and exit dropped from 25.2% to 7.7%.

Table 38: Goal Assessment for Hennepin County Veterans Court Evaluation

Goal Assessment Rationale

1) Facilitate participant sobriety Needs improvement

Over two-fifths of all participants never drug tested; No random drug testing regimen; Should continue testing after program completion when participants are on administrative probation

2) Increase compliance with treatment and other court ordered conditions

Meets goal

Most participants have successful treatment outcomes; Participants leave program with additional diagnoses to manage; Warrants are more detrimental to program success than in-program charges/convictions

3) Improve access to VA benefits and services Meets goal Program has been highly successful

increasing participants’ access to benefits

4) Improve family relationship and social support connections

Partially meets goal

Participants tend to enhance prosocial relationships and activities during participation. Program needs to expand the rebuilt mentor program as soon as possible.

5) Improve life stability Partially meets goal Need additional attention to improve housing stability and employment opportunities

6) Reduce criminal recidivism Needs improvement

Veterans Court participants did not recidivate less than a matched comparison group of veteran offenders, though they did have fewer total new convictions; Veterans Court graduates reoffended less than a matched comparison group.

69


Having representatives from the VJO and the VBA on the Veterans Court team appears to be

instrumental in helping veterans receive the benefits they deserve.

Moving to the goal of enhancing family relationships and improving social supports, the number of

prosocial activities and relationships tended to increase over the course of program participation—

especially for graduates—suggesting many individuals are enhancing the array of positive social

supports due to the program. In addition, participants paired with a mentor succeeded at higher rates

than individuals without a mentor. Despite the success of the mentor program, which assisted the

Veterans Court program in its goal of improving family relationships and social support connections,

the mentor program did not function between 2014 and 2015, and is only getting up and running

again in 2016. Despite the difficulties sustaining the mentor program, the Hennepin County Veterans

Court program was quite successful overall in the goal of improving family relationship and social

support connections.

The program was largely successful in the goal of improving life stability. Many participants increased

their levels of education, housing stability, and employment, while many more maintained high levels

of life stability throughout the program. Nevertheless, there is still work to do, especially in the areas

of housing, where 14 individuals were homeless upon graduation. Despite new national and state

programs intended to help end homelessness among the veteran population, the program must

provide veterans in the criminal justice system additional or different assistance to ensure all

participants have stable housing. In addition, 24 participants left the program unemployed. Finding

suitable work can be difficult, especially for individuals who have had contact with the criminal justice

system, but with the Minnesota Department of Employment and Economic Development (DEED)

at the table, it should be easier to find stable and suitable employment for all interested participants.

Finally, Veterans Court participants do not recidivate less in the two years after they leave the program

than the comparison group of US Armed Forces veterans who did not participate in the program.

However, this analysis demonstrated that the Veterans Court cohort recidivated in different ways than

the comparison group. First of all, even though the proportion of Veterans Court participants who

recidivated was not statistically different than the proportion of the comparison group who committed

new criminal offenses, Veterans Court graduates were convicted of fewer new offenses than the

comparison group during the two year window. Second, the Veterans Court participants who

70


recidivated did so quickly after exiting the program: more than half recidivated within several months.

Following the recommendation of a graduated drawdown in supervision after leaving the program

could help further reduce reoffending and help the program fully meet this goal going forward. Finally,

Veterans Court graduates did recidivate significantly less than a match comparison group of justice-

involved veterans who were statistically identical to program graduates. This suggests that program

graduates maintain the behavioral changes developed during Veterans Court.

Despite the fact that this evaluation shows that the Hennepin County Veterans Court did not meet all

of its goals, it bears mentioning that in the final analysis the program is by no means failing. Even in

the areas where the program needs improvement, there are tangible successes and highlights, and low-

hanging fruit that provide opportunities for change. Indeed, the deficiencies with the goal of

enhancing participant sobriety can improve by moving closer to the best practices for drug testing as

articulated by the NADCP. Furthermore, rebuilding the mentorship program could enhance the

graduation rate and help reduce recidivism. Thus, identifying and codifying changes to enhance

program outcomes is well within the grasp of the Steering Committee.

In several years, the Hennepin County District Court Research Division will evaluate the Veterans

Court program again, likely examining participants active between 2014 and 2016. Historically, the

retrospective nature of evaluations exposes data collection and quality issues that are difficult to

reconcile; at best, filling in these gaps requires time and energy, at worst, these difficulties result in

missing data. However, the improvements made to Treatment Court data collection and data

management since 2014—principally the creation of the Hennepin County District Court Treatment

Court Database and the Research Division team member hired to manage Treatment Court data—

should expedite this subsequent evaluation while minimizing missing data issues. It is hopeful that the

next Veterans Court evaluation will demonstrate even more program successes.

71


REFERENCES

ACLU. “Picking Up the Pieces: A Minneapolis Case Study.” New York, NY, 2014:

https://www.aclu.org/feature/picking-pieces. Baldwin, Julie Marie. “Investigating the Programmatic Attack: A National Survey of Veterans

Treatment Courts.” Journal of Criminal Law and Criminology, 105, 2016: 705-750. Caron, Anne. “Fourth Judicial District Veterans Court – Two Year Review: July 2010 – June 2012.”

Minneapolis, MN: Fourth Judicial District Research Division, 2013. O’Brien, Christine Neylon and Jonathan J. Darrow. “Adverse Employment Consequences Triggered

by Criminal Convictions: Recent Cases Interpret State Statutes Prohibiting Discrimination.” Wake Forest Law Review, 27, 2007: 991-1028.

Department of Defense, “2014 Demographics, Profile of the Military Community.” Washington, D.C: United States Department of Defense, 2014.

Hartley, Richard D., and Julie Marie Baldwin. “Waging War on Recidivism among Justice-Involved Veterans: An Impact Evaluation of a Large Urban Veterans Treatment Court.” Criminal Justice Policy Review, 2016: 1-27.

Johnson, Matthew. “Adult Dispositional Analysis.” Minneapolis, MN: Fourth Judicial District Research Division, 2015.

Department of Veterans Affairs. “Under Secretary for Health’s Information Letter.” Washington, D.C.: Department of Veterans Affairs, 2009.

Loeber, Rolf, and Rebecca Stallings, “Modeling the Impact of Interventions on Local Indicators of Offending, Victimization, and Incarceration,” in Young Homicide Offenders and Victims: Risk Factors, Prediction, and Prevention from Childhood, eds. Rolf Loeber and David P. Farrington, New York: Springer, 2011: 137-152.

Mitchell, Ojmarrh, David B. Wilson, Amy Eggers, and Doris L. MacKenzie. “Assessing the effectiveness of drug courts on recidivism: A meta-analytic review of traditional and non-traditional drug courts.” Journal of Criminal Justice 40, 2012: 60-71.

National Association of Drug Court Professionals. “Adult Drug Court Best Practice Standards, Vol II.” Alexandria, VA: 2015.

National Association of Drug Court Professionals. “Adult Drug Court Best Practice Standards, Vol I.” Alexandria, VA: 2013.

Wilder Research. “Homeless Veterans in Minnesota 2015: Statewide Survey of People Experiencing Homelessness.” St. Paul, MN: 2016.